applied sciences

Article Development of an Unmanned Surface Vehicle for the Emergency Response Mission of the ‘Sanchi’ Collision and Explosion Accident

Huayan Pu, Yuan Liu , Jun Luo, Shaorong Xie, Yan Peng *, Yi Yang, Yang Yang, Xiaomao Li, Zhou Su, Shouwei Gao, Wenyun Shao, Chuang Zhu, Jun Ke, Jianxiang Cui and Dong Qu

School of Mechatronic Engineering and Automation, University, Shanghai 200444, China; [email protected] (H.P.); [email protected] (Y.L.); [email protected] (J.L.); [email protected] (S.X.); [email protected] (Y.Y.); [email protected] (Y.Y.); [email protected] (X.L.); [email protected] (Z.S.); [email protected] (S.G.); [email protected] (W.S.); [email protected] (C.Z.); [email protected] (J.K.); [email protected] (J.C.); [email protected] (D.Q.) * Correspondence: [email protected]

 Received: 19 March 2020; Accepted: 10 April 2020; Published: 14 April 2020 

Abstract: Unmanned surface vehicles (USVs) as unmanned intelligent devices can replace humans to perform missions more efficiently and safely in dangerous areas. However, due to the complex navigation environment and special mission requirements, USVs face many challenges in emergency response missions for marine oil spill accidents. To solve these challenges in the emergency response mission of the ‘Sanchi’ oil tanker collision and explosion accident, we designed and deployed an USV to perform the missions of real-time scanning and water sampling in the shipwreck waters. Compared with the previous USVs, our USV owned the following characteristics: Firstly, the improved navigation control algorithms (path following and collision avoidance) can provide high navigation accuracy while ensuring navigation safety; Secondly, an improved launch and recovery system (LARS) enabled the USV to be quickly deployed and recovered in the mission area; Thirdly, a new sampling system was specially designed for the USV. Our USV completed the missions successfully, not only providing a lot of information for rescuers but also offering a scientific basis for follow-up work.

Keywords: unmanned surface vehicle; marine oil spill; emergency response

1. Introduction On 6 January 2018, the Iranian oil tanker ‘Sanchi’ carrying 111,300 tons of condensate collided with the Hong Kong ‘Changfeng Crystal’ in the East China sea, at approximately 160 nautical miles east of Shanghai. After the collision, the ‘Sanchi’ burned violently and exploded on 14 January. The ‘Sanchi’ sank 4 h after the explosion. Three people died and 29 people were missing in the accident, and oil leaked from the ‘Sanchi’ polluted 10 km2 of sea area. After the accident, the State Oceanic Administration of China, the Maritime Search and Rescue Center of China, and the Maritime Search and Rescue Center of Shanghai launched the emergency response mission for the accident. Figure1 is the scene of the accident. In this accident, the ‘Sanchi’ oil tanker carried a total of 113,000 tons of condensate. Condensate is a highly toxic and volatile chemical substance, and there is a risk of deflagration at any time after it leaks. Therefore, it is too dangerous for humans to enter the core area of the accident to perform a mission. We urgently needed a mobile unmanned platform to replace humans in the core area of the accident to obtain the necessary information, such as the submarine topography, the state of the shipwreck, the location of oil spills, and the situation of water pollution.

Appl. Sci. 2020, 10, 2704; doi:10.3390/app10082704 www.mdpi.com/journal/applsci Appl. Sci. 2020, 10, 2704 2 of 21 Appl. Sci. 2020, 10, x FOR PEER REVIEW 2 of 21

FigureFigure 1.1. Accident scene.

UnmannedIn this accident, surface the vehicles ‘Sanchi’ (USVs) oil tanker are carried the suitable a total platform of 113,000 for antons emergency of condensate. response Condensate mission. Unmannedis a highly toxic meant and it couldvolatile be chemical deployed substance, for a longer and time, there therefore is a risk wideningof deflagration the operational at any time window. after it Mostleaks. importantly, Therefore, it the is usetoo ofdangerous the USVs for resolved humans the to security enter the challenge core area of surveying of the accident in dangerous to perform areas. a Inmission. recent years,We urgently USVs have needed been a widelymobile researchedunmanned and platform applied to in replace a lot of humans fields [1 ].in Somethe core new area USVs of arethe constantlyaccident to developed, obtain the such necessary as ‘Spartan information, Scout’ designed such as by the the submarine USV team intopography, San Diego [the2], ‘Nighthawk’state of the fromshipwreck, AAC the (Accurate location Automation of oil spills, and Corp) the [ 3situation], and a of range water of pollution. USVs from ASV(The Company of AutonomousUnmanned Surface surface Vechile), vehicles such as (USVs) the USVs are that the called suitable ‘C-WORKER’ platform and for ‘C-TARGET’ an emergency [4]. Although response USVsmission. have Unmanned many applications meant it at could this stage, be deployed they are limited for a longer to carrying time, out therefore missions widening for scientific the research,operational such window. as environmental Most importantly, monitoring the [ use5–7 ], of environmental the USVs resolved sampling the [ security8,9], communication challenge of platformsurveying [10 in, 11dangerous], and harbor areas. protection In recent andyears, patrol USVs [12 have–15]. been However, widely emergency researched response and applied mission in a forlot disasterof fields or[1]. accident Some new are USVs performed are constantly by AUV(Autonomous developed, such Underwater as ‘Spartan Vehicle) Scout’ ordesigned ROV(Remote by the OperatedUSV team Vehicle),in San Diego for example, [2], ‘Nighthawk’ Kukulya, from A.L. AAC et al. (Accurate developed Automation an AUV-based Corp)[3], approach and a inspiredrange of byUSVs an existing from ASV(The small, long-range Company system, of Autonomous called the Surface Tethys Long-Range Vechile), such AUV as (LRAUV),the USVs inthat order called to support‘C-WORKER’ the Arctic and Doman ‘C-TARGET’ Awareness [4]. Although Center (ADAC) USVs forhave spill many preparedness applications [16]. at Valentine, this stage, M.M. they et are al. usedlimited remote-controlled to carrying out vehicles missions to for quantify scientific large research, animals suchat five as study environmental sites to determine monitoring thee ff [5–7],ects ofenvironmental the spill on deep-water sampling [8–9], animals communication [17].The only platform application [10–11], of USVs and for harbor disaster protection was responding and patrol to Hurricane[12–15]. However, Wilma. During emergency the rescue response mission, mission the Center for disaster for Robot-Assisted or accident Search are performed and Rescue by at theAUV(Autonomous University of South Underwater Florida used Vehicle) a USV or thatROV(Remote called ‘AEOS-1’ Operated to inspect Vehicle), docks for example, and seawalls Kukulya then bridgesA L et al. [18 ].developed an AUV-based approach inspired by an existing small, long-range system, calledIn the this Tethys emergency Long-Range response AUV mission, (LRAUV), we deployed in order a newlyto support designed the Arctic USV. ComparedDoman Awareness with the previousCenter (ADAC) USVs, (1) for the spill improved preparedness[16]. navigation control Valentine algorithms M M et (path al. used following remote-controlled and collision avoidance)vehicles to canquantify provide large high animals navigation at five accuracy study while sites ensuring to determine the navigation the effects safety; of the (2) spillthe improved on deep-water LARS (Launchanimals[17].The And Recovery only application System) enabled of USVs the for USV disaster to be quicklywas responding deployed to and Hurricane recovered Wilma. in the missionDuring area;the rescue and (3) mission, a new pump the Center sampling forsystem, Robot-Assisted which is small Search in and size andRescue light at in the weight, University can be ofinstalled South orFlorida removed used at a anyUSV time. that called ‘AEOS-1′ to inspect docks and seawalls then bridges [18]. TheIn this rest emergency of this paper response is organized mission, according we deployed to the structure a newlybelow. designed In Section USV. Compared2, we introduce with thethe systemprevious architecture USVs, (1) of the the improvedUSV. The approach navigation is proposed control algorithms in Section3 . (path In Section following4, we introduce and collision the missionavoidance) process can and provide analyze high the navigation mission data. accuracy Finally, while the conclusion ensuring is the given navigation in Section safety;5. (2) the improved LARS (Launch And Recovery System) enabled the USV to be quickly deployed and recovered in the mission area; and (3) a new pump sampling system, which is small in size and light in weight, can be installed or removed at any time. Appl.Appl. Sci. Sci. 2020 2020, ,10 10, ,x x FOR FOR PEER PEER REVIEW REVIEW 3 3of of 21 21

TheThe rest rest of of this this paper paper is is organized organized according according to to the the structure structure below. below. In In Section Section 2, 2, we we introduce introduce theAppl.the system system Sci. 2020 architecture ,architecture10, 2704 of of the the USV. USV. The The approach approach is is proposed proposed in in Section Section 3. 3. In In Section Section 4, 4, we we introduce introduce3 of 21 thethe mission mission process process and and analyze analyze the the mission mission data. data. Finally, Finally, the the conclusion conclusion is is given given in in Section Section 5. 5.

2.2. USV USVUSV System SystemSystem Structure StructureStructure OurOur USV USV USV consisted consisted consisted of the of of followingthe the following following parts: parts: Hull, parts: control Hull, Hull, module,control control locomotionmodule, module, locomotion locomotion module, navigation module, module, navigationmodule,navigation and module, module, mission and and module. mission mission Each module. module. of the Each partsEach of willof the the be parts parts discussed will will be be in discussed discussed turn. Figure in in turn.2 turn. is an Figure Figure overview 2 2 is is an ofan overviewtheoverview USV’s of architecture.of the the USV’s USV’s architecture. architecture. USVUSV

NavigationNavigation module module DiselDisel EngineEngine

Forward LiDARLiDAR Forward Radar CCDCCD INS GPSGPS lookinglooking sonar sonar Radar CammerCammer INS Water-jetWater-jet propulsionpropulsion

AcquisitionAcquisition construction construction EnvironmentEnvironment information information

BottomBottom control control box box DieselDiesel Main controller MotionMotion MotionMotion ControlControl Main controller informationinformation constructionconstruction generatorgenerator modulemodule

AcquisitionAcquisition construction construction MissionMission information information FuelFuel tank tank

Launch and Multi-beamMulti- Single-Single- ADCPADCP SASSAS SamplingSampling Launch and bathymetrybathymetry SideSide scan scan recoveryrecovery beambeam sonarsonar devicedevice device bathymetrybathymetry device

BatteryBattery ScanningScanning devicedevice LocomotionLocomotion MissionMission module module modulemodule FigureFigureFigure 2. 2.2. Unmanned UnmannedUnmanned surface surfacesurface vehicle vehiclevehicle (USV) (USV)(USV) architecture architecturearchitecture overview. overview.overview. 2.1. Hull Design 2.1.2.1. Hull Hull Design Design The USV weighed 2300 kg and was 6.28 m 0.98 m 0.32 m (length width height) with a TheThe USV USV weighed weighed 2300 2300 kg kg and and was was 6.28 6.28 m m ×× × 0.98 0.98 m m × × 0.32 0.32 m m (length (length × × width width × × height) height) with with a a maximum payload of 1 ton. The hull was divided into three compartments. Forward-looking sonar maximummaximum payload payload of of 1 1 ton. ton. The The hull hull was was divided divided into into three three compartments. compartments. Forward-looking Forward-looking sonar sonar was installed at the front section. In consideration of vibration problems, an instrument cabinet which waswas installed installed at at the the front front section. section. In In consideration consideration of of vibration vibration problems, problems, an an instrument instrument cabinet cabinet whichelasticallywhich elastically elastically connected connected connected with the with with deck the the was deck deck located was was inlocated located the middle in in the the section. middle middle On section. section. the other On On hand,the the other other a circulating hand, hand, a a cooling system can prevent the instrument cabinet from overheating. Inboard diesel engine, fuel tank, circulatingcirculating cooling cooling system system can can prevent prevent the the instrument instrument cabinet cabinet from from overheating. overheating. Inboard Inboard diesel diesel engine,battery,engine, fuel andfuel tank, water-jettank, battery, battery, propulsion and and water-jet water-jet were installedpropulsion propulsion in werethe were tail installed installed section. in in The the the installationtail tail section. section. locationThe The installation installation of each device is shown in Figure3. locationlocation of of each each device device is is shown shown in in Figure Figure 3. 3.

Figure 3. Device installation location. Appl. Sci. 2020, 10, x FOR PEER REVIEW 4 of 21

Figure 3. Device installation location. Appl. Sci. 2020, 10, 2704 4 of 21 2.2. Control Module 2.2. ControlThe Module control module can collect sensor information in real-time, control the USV’s movement, manageThe control the modulemission can device, collect and sensor realize information the independent in real-time, work control of the USV. the USV’s The control movement, module manageconsisted the mission of an device,industrial and personal realize the computer independent (IPC) work and ofa bottom the USV. control The control box which module was consisted equivalent of anto industrial the brain personal of the USV. computer The environmental (IPC) and a bottom information, controlbox the whichontology was information, equivalent to and the the brain target of theinformation USV. The environmentalcollected by the information,sensors werethe transmitted ontology to information, the control module. and the targetThe control information algorithms collectedpre-embedded by the sensors in the were IPC transmitted processed to this the information control module. and Thethen control different algorithms control pre-embedded instructions were in thecalculated IPC processed and transmitted this information to other andmodules. then different control instructions were calculated and transmittedThe to key other of modules.the control module was the IPC interfaced with a serial extension board (Digital SignalThe key Processor) of the control and a module network was switch. the IPC The interfaced general IPC with from a serial YanHua’s extension APAX board series (Digital was Signal used as a Processor)hardware and aplatform network and switch. the Theembedded general control IPC from system YanHua’s based APAX on Linux series waswas useddeveloped. as a hardware Since data platformcould and be collected, the embedded processed, control and system stored based independently on Linux wasby the developed. IPC, the loss Since of datathe data could could be be collected,effectively processed, avoided and in storedthe case independently of communication by the interruption IPC, the loss or ofindirect/incomplete the data could be transmission. effectively avoided in the case of communication interruption or indirect/incomplete transmission. 2.3. Locomotion Module 2.3. LocomotionThe locomotion Module module was the execution module of our USV. After receiving the control instructions,The locomotion it drove module the USV was to the perform execution various module actions. of our Diesel USV. engine, After diesel receiving generator, the control water-jet instructions,propeller, it fueldrove tank, the USV and battery to perform constituted various actions.the locomotion Diesel engine, module. diesel It is generator, worth mentioning water-jet that propeller,traditional fuel tank,propeller and propulsion battery constituted was replaced the locomotion by a water-jet module. propulsion. It is worth Bettermentioning working ability that and traditionalmaneuverability propeller propulsionin the shallow was water, replaced higher by a thrust, water-jet and propulsion. lower noise Better were workingthe obvious ability advantages and maneuverabilityof water-jet in propulsion the shallow technology water, higher over thrust, propeller and lower propulsion noise were [19–20]. the obvious The locomotion advantages module of water-jetensured propulsion that the technologymaximum speed over propeller of the USV propulsion was no less [19 ,20than]. The13 kn(Knot), locomotion the module cruising ensured speed was thatabout the maximum 10 kn, and speed the of ranges the USV of wasworking no less speed than were 13 kn(Knot), 4 to 8 kn. the The cruising USV speedcould wasrun aboutfor about 10 kn, 5 h at andcruising the ranges speed of working and 12 h speed at working were 4speed. to 8 kn. The USV could run for about 5 h at cruising speed and 12 h at working speed. 2.4. Navigation Module 2.4. Navigation Module The navigation module received data from sensors such as GPS(Global Positioning System), INS(InertialThe navigation Navigation module received System), data compass, fromsensors forward such looking as GPS(Global sonar, radar, Positioning and LiDAR(Light System), INS(InertialDetection Navigation and Ranging), System), etc. compass, After getting forward the looking sensor sonar, data, radar, the navigation and LiDAR(Light module Detection analyzed the andUSV's Ranging), position, etc. After speed, getting heading, the sensor and data, obstacle the navigation information, module and gave analyzed the USV'sthe USV’s next position, speed and speed,heading heading, combined and obstacle with the information, mission requirements. and gave the USV’sIf obstacles next speedexisted, and the heading speed and combined heading with had to the missionbe adjusted requirements. through If the obstacles corresponding existed, the collision speed and avoidance heading algorithm. had to be adjusted The architecture through the of the correspondingnavigation collisionmodule is avoidance shown in algorithm. Figure 4. The architecture of the navigation module is shown in Figure4.

Navigation Sensor data module

Obstacle Location Mission Information information requirements

Obstacle Path detection following algorithm

Initial speed Initial heading

Collision Speed avoidance algorithm Heading

FigureFigure 4. Navigation 4. Navigation module. module. Appl. Sci. 2020, 10, 2704 5 of 21 Appl. Sci. 2020, 10, x FOR PEER REVIEW 5 of 21

2.5. Mission Mission Module Module The mission mission module module was was divided divided into into three three systems: systems: Scanning, Scanning, sampling, sampling, and LARS. and The LARS. scanning The scanningsystem mainly system used mainly the used device the todevice realize to therealize underwater the underwater exploration. exploration. The sampling The sampling system system was wasdesigned designed to collect to collect water water samples samples in real in real time. time. LARS LARS was was used used to deploy to deploy and and recover recover the the USV USV in the in themission mission area. area.

3. Approach Approach

3.1. Path Following Algorithm 3.1. Path Following Algorithm A straight path following algorithm based on the predicted position control was adopted in our A straight path following algorithm based on the predicted position control was adopted in our USV. The core idea of the algorithm was to calculate the d (the USV’s predicted lateral offset). If d was USV. The core idea of the algorithm was to calculate the0 ’(the USV’s predicted lateral offset).0 If ’ larger, the angle which made the USV return to the planned path was also greater. was larger, the angle which made the USV return to the planned path was also greater. The schematic diagram of the algorithm is shown in Figure5. We defined some variables, where The schematic diagram of the algorithm is shown in Figure 5. We defined some variables, Yaw and Angel represent the course angle and turning angle required for the USV to return to the where and represent the course angle and turning angle required for the USV to return planned path, and d and d represent the actual lateral offset and predicted lateral offset, respectively. to the planned path, and 0 and ’ represent the actual lateral offset and predicted lateral offset, The calculation formula of d is Equation (3). The calculation formula of d is Equation (2), where the respectively. The calculation formula of is Equation (3). The calculation0 formula of ’ is Equation linear speed and angular speed of the USV were set to Vusv and ωusv. Yaw and Angel are calculated (2), where the linear speed and angular speed of the USV were set to and ωusv. and as Equation (1), where TogestDist and ToDestAng are the distance and angle from the USV to the are calculated as Equation (1), where and are the distance and angle from terminating waypoint, and is the heading angle of the USV. PathAng is the angle of the path, which is the USV to the terminatingϕ waypoint, and is the heading angle of the USV. ℎ is the angle ToDestDist ToDestAng ofcalculated the path, from which the start is calculated and end waypoint from the of startthe path. and end waypointand of the path.are calculated based and on the USV’s state (position, heading angle, the latitude and longitude information to the terminating are calculated based on the USV’s state (position, heading angle, the latitude and waypoint, etc.). Parameters T1 and T2 are called time-lag constant and time-lag slope, respectively. The longitude information to the terminating waypoint, etc.). Parameters and are called time-lag f (x) constantis a saturationand time-lag function, slope, inrespectively. order to modify The the() distance is a saturation conversion function, coeffi incient orderχd intoto modify a variable the form. Therefore, both the real time and smoothness of the USV movement are considered. distance conversion coefficient into a variable form. Therefore, both the real time and smoothness of the USV movement are considered. Yaw = ToDestAng + f (d0) χ Angel = ToDestAng + f (d0) χ ϕ (1) × d × d − = + (’) × = + (’) × − (1) d0 = d + Vusv T1 sin(PathAng ωusv T2) (2) ’ = + × × ( ℎ× −× × ) − × (2) d = ToDestDist sin (PathAng ToDestAng) (3) = × × (ℎ× −×−) (3) 60.0x 60.0 f (x) =   (4) ( ) 0.5( x 10) = .(||) x 1 + e (4) ||(1 + ) | | − | |−

Figure 5. Path-following algorithm based on predicted position control. Figure 5. Path-following algorithm based on predicted position control.

In order to calculate and , , , and are the three parameters that must be calculated, so we focused on how to determine these parameters. Appl. Sci. 2020, 10, 2704 6 of 21

In order to calculate Yaw and Angel, T1, T2, and χd are the three parameters that must be calculated, so we focused on how to determine these parameters. Time-lag constant T1: Parameter T1 acting on Vusv was mainly used to calculate d0. According to the results of the sea trials, when the USV had a set throttle value of 500, T1 = 5 could make the algorithm achieve a higher control accuracy. When the throttle value increased, the T1 decreased accordingly. Table1 shows the T1 for each throttle setting.

Table 1. Selection of T1 for each throttle value. Throttle Value 300 400 500 600 700 800 Time delay constant T1 6 5.5 5 5 3.5 2

Time-lag slope T2: Parameter T2 acting on ωusv was related to the sea states during navigation. According to the results of the sea trials, it was appropriate to set the T2 to 5 when the sea states were 1–2. When the sea state was 3, it was appropriate to set the T2 to 5.5 or more. This is because the USV was more difficult to turn in poor sea states, therefore, it was necessary to give a long forecast value when calculating d0. Distance conversion coefficient χd: χd limits the values of Yaw and Angel. This parameter was generally set to 2~3. According to the results of the sea trials, χd = 3 was a suitable value. If set to 3, the maximum control amount of the angle was 60◦.

3.2. Collision Avoidance Algorithm The control of the collision avoidance for USVs is generally divided into two layers [21]. The first layer is global collision avoidance, which is to construct a proper environment model based on the information of known obstacles or danger areas and then plan a safe path. The second layer is the short-range real-time collision avoidance based on the real-time sensor information, which enables USVs to quickly avoid the obstacle that appears suddenly in the surrounding environment. Therefore, short-range real-time collision avoidance is the last line of defense to ensure the safety of USVs. Velocity obstacle (VO) algorithm is a kind of short-range real-time collision avoidance algorithm commonly used in USVs. However, the traditional VO algorithm does not consider the influence of the kinematic performance of USVs and the error of obstacle movement information, nor does it specify when to start collision avoidance and when to complete. Therefore, in order to overcome these problems, we combined a dynamic window algorithm with elliptic VO algorithm, gave the judgment conditions for when to start and end collision avoidance, and used the virtual obstacle method to reduce the influence of the error of obstacle movement information.

3.2.1. Elliptic VO Algorithm The basic principle of VO algorithm was introduced in detail in [22], and the key difference between the elliptic VO algorithm and the VO algorithm is the process of solving the tangent. In the VO algorithm, if you want to judge whether a USV will collide with an obstacle, you must require two tangent lines of the obstacle relative to the USV. As shown in Figure6, in order to find the tangent lines of elliptical obstacle conveniently, a body-fixed coordinate system and an obstacle coordinate system was established. The xOy is the body-fixed coordinate system of the USV and x0O0 y0 is the obstacle coordinate system. The red ellipse is an obstacle and the green ellipse is an USV, which was simplified as a particle. Point O is the center of the USV. O0 is the center of the obstacle. If the coordinates of the USV in the body-fixed coordinate T system are USVboat = [0, 0] , the coordinates of the USV in the obstacle coordinate system are:

1 USV = R− (USV C) (5) obs boat − Appl. Sci. 2020, 10, 2704 7 of 21 where C is the coordinate of the center point of the obstacle in the body-fixed coordinate system and R is the rotation matrix. " # Appl. Sci. 2020, 10, x FOR PEER REVIEW cosθ sinθ 7 of 21 R = (6) sinθ cosθ − When the coordinates of the USV in the obstacle coordinate system are obtained, the tangent When the coordinates of the USV in the obstacle coordinate system are obtained, the tangent coordinate (, ) in the obstacle coordinate system can be obtained by Equation (7) (1 and coordinate (xT, yT) in the obstacle coordinate system can be obtained by Equation (7) (T1 and T2 2 in Figure 6): obs obs in Figure6):  a2 y2  T T + = 1  2 + 2 = 1  a b (7)  x m y n (7)  T + T = 1, + = 1, a2 b2 Then, T1obs, T2obs is converted into the body-fixed coordinate system by using Equation (8): Then, 1, 2 is converted into the body-fixed coordinate system by using Equation (8): Tboat = RTobs + C (8) = + (8)

FigureFigure 6. 6. EllipseEllipse tangents. tangents. 3.2.2. VO Algorithm Based on Dynamic Window Algorithm 3.2.2. VO Algorithm Based on Dynamic Window Algorithm After calculating the colliding VO sets by elliptic VO algorithm, the data in the non-VO sets were After calculating the colliding VO sets by elliptic VO algorithm, the data in the non-VO sets the optional collision avoidance velocity vectors for the USV. However, due to the constraint of the were the optional collision avoidance velocity vectors for the USV. However, due to the constraint of dynamic performance of the USV, many velocity vectors in the non-VO sets were inaccessible to the the dynamic performance of the USV, many velocity vectors in the non-VO sets were inaccessible to USV. The dynamic window algorithm considered the kinematic performance of the USV [20], and only the USV. The dynamic window algorithm considered the kinematic performance of the USV [20], calculated the moving velocity that the USV can reach within a given time window ∆t, that is, the and only calculated the moving velocity that the USV can reach within a given time window ∆, that velocity window vd: is, the velocity window : n h . . io vd = v v vc v∆t, vc + v∆t (9) = | ∈ − ∆, + ∈∆− (9) and the angular velocity that can be achieved, that is, the angular velocity window ωd: and the angular velocity that can be achieved, that is, the angular velocity window : n h . . io = | ∈ ω−d = ∆ω, ω + ω c∆ω ∆ t , ω c + ω ∆ t (10)(10) ∈ −

In Equation (9), is the current moving speed of the USV and. is the acceleration of the In Equation (9), vc is the current moving speed of the USV and v is the acceleration of the USV. USV. In Equation (10), is the current angular velocity of the. USV and is the angular In Equation (10), ω is the current angular velocity of the USV and ω is the angular acceleration of the acceleration of the USV. According to Equation (11), the heading angle that the USV can get within USV. According to Equation (11), the heading angle that the USV can get within ∆t can be calculated: ∆ can be calculated:

  . .  1 1 2 1 1 2 = θ∈ = θ+ θ ∆θ−+ω ∆c∆t, +ω∆t ∆, θ++ ω c∆∆t + ω ∆ t (11)(11) d ∈ h 2 − 2 h 2 2 where is the current heading of the USV. where θh is the current heading of the USV. Through the Equations (9) and (11), and can be determined, but the elements in sets of Through the Equations (9) and (11), vd and θd can be determined, but the elements in sets of vd and and are continuous, which is not conducive to calculation in engineering. Therefore, is θd are continuous, which is not conducive to calculation in engineering. Therefore, vd is discretized discretized into speeds and is discretized into heading directions. Each discretized into M speeds and θd is discretized into N heading directions. Each discretized velocity vi and heading velocity and heading form a velocity vector (, ), and then there are × velocity vectors. The set of velocity vectors is called reachable velocity (RV). The velocity vectors satisfying Equation (9) in the RV set will collide with the obstacle. Excluding these velocity vectors from the RV set can obtain a safe set of velocity vectors, called reachable avoidance velocity (RAV): = | ∈ , ∉ (12) Appl. Sci. 2020, 10, 2704 8 of 21

θ form a velocity vector (v , θ ), and then there are M N velocity vectors. The set of velocity vectors i i i × is called reachable velocity (RV). The velocity vectors satisfying Equation (9) in the RV set will collide with the obstacle. Excluding these velocity vectors from the RV set can obtain a safe set of velocity vectors, called reachable avoidance velocity (RAV):

RAV = V V RV, V < VO (12) Appl. Sci. 2020, 10, x FOR PEER REVIEW { | ∈ } 8 of 21 As shown in Figure7, the red area is VO area, the intersection of RV and VO is the velocity vector As shown in Figure 7, the red area is VO area, the intersection of RV and VO is the velocity where collision will occur, and the RAV part is the safe velocity vector. vector where collision will occur, and the RAV part is the safe velocity vector.

FigureFigure 7. 7. RV(ReachableRV(Reachable Velocity)and Velocity)and RAV (Reachable AvoidanceAvoidance Velocity).Velocity).

After obtaining the RAV, itit waswas necessarynecessary toto selectselect thethe appropriateappropriate velocity vector in the RAV according toto thethe mission mission requirements. requirements. In In di ff differenterent missions, missions, there there are are diff erentdifferent selection selection criteria, criteria, and theand following the following evaluation evaluation formula formula was usedwas used to select to select in this in mission:this mission: 1 1 = G = (13)(13) ‖( cosij − cos , sin − sin )‖ (vc cos θ v cos θ , vc sin θ v sin θ ) || h − i i h − i i || where vc is the current speed of the USV, θ is the current heading of the USV, (vi, θi) is a certain where is the current speed of the USV, h is the current heading of the USV, ( , ) is a certain velocity vector in RAV, and (v , θ ), which can maximize G , is taken as the velocity vector for the USV velocity vector in RAV, and i(i, ), which can maximizeij , is taken as the velocity vector for the toUSV avoid to avoid the obstacle. the obstacle. Once Once a velocity a velocity vector vector is selected, is selected, the USV the will USV always will always travel alongtravel thisalong speed this vectorspeed vector until there until are there new are dangerous new dangerous situations situations or the obstacleor the obstacle is completely is completely avoided. avoided. 3.2.3. Starting Collision Avoidance 3.2.3. Starting Collision Avoidance When judging whether the USV will collide with an obstacle, the distance between the USV and When judging whether the USV will collide with an obstacle, the distance between the USV and the obstacle is the smallest at the collision time, and their respective positions are called CPA(Close the obstacle is the smallest at the collision time, and their respective positions are called CPA(Close Point Approach). The time required for the USV to reach its CPA point is called tCPA. As shown Point Approach). The time required for the USV to reach its CPA point is called . As shown in in Figure8, the red and green solid ellipses are the positions of the obstacle and USV at the time t0, Figure 8, the red and green solid ellipses are the positions of the obstacle and USV at the time , Vobs and Vusv are the velocity vectors of the obstacle and USV at time t0, the red and green dashed and are the velocity vectors of the obstacle and USV at time , the red and green dashed ellipses are the positions of the obstacle and USV at time tCPA, and the CPAobs and CPAUSV points are ellipses are the positions of the obstacle and USV at time , and the and points the positions of the center points of the obstacle and USV at time tCPA. The tCPA is calculated from are the positions of the center points of the obstacle and USV at time . The is calculated Equation (14): from Equation (14): (PUSV P )(VUSV V ) t = − obs − obs (14) ( −CPA )( − ) 2 (VUSV Vobs) = || − || (14) ‖( − ‖

Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 21

As shown in Figure 7, the red area is VO area, the intersection of RV and VO is the velocity vector where collision will occur, and the RAV part is the safe velocity vector.

Figure 7. RV(Reachable Velocity)and RAV (Reachable Avoidance Velocity).

After obtaining the RAV, it was necessary to select the appropriate velocity vector in the RAV according to the mission requirements. In different missions, there are different selection criteria, and the following evaluation formula was used to select in this mission: 1 = (13) ‖( cos − cos , sin − sin )‖

where is the current speed of the USV, is the current heading of the USV, ( , ) is a certain velocity vector in RAV, and ( , ), which can maximize , is taken as the velocity vector for the USV to avoid the obstacle. Once a velocity vector is selected, the USV will always travel along this speed vector until there are new dangerous situations or the obstacle is completely avoided.

3.2.3. Starting Collision Avoidance When judging whether the USV will collide with an obstacle, the distance between the USV and the obstacle is the smallest at the collision time, and their respective positions are called CPA(Close

Point Approach). The time required for the USV to reach its CPA point is called . As shown in Figure 8, the red and green solid ellipses are the positions of the obstacle and USV at the time , and are the velocity vectors of the obstacle and USV at time , the red and green dashed ellipses are the positions of the obstacle and USV at time , and the and points are the positions of the center points of the obstacle and USV at time . The is calculated from Equation (14):

( − )( − ) = (14) Appl. Sci. 2020, 10, 2704 ‖( − ‖ 9 of 21

Figure 8. Close point approach and start avoidance point.

Assuming that the speed of the USV is kept constant, i.e., the USV can turn at a constant angular acceleration. In Figure8, in the process of sailing to CPAUSV point, the USV turns left with a constant angular acceleration at point B, which is tangent to the obstacle at point PL and just along the path B PL. The time required for the USV to sail from point B to point PL is tL. Then the USV starts to turn − left to avoid the obstacle at the time tCPA = tL, which is just tangent to the obstacle at point PL. PL is calculated from Equation (15):

 R t    x = x + L v cos θ + ω t + 1 αt2 dt  B 0 0 0 2  R tL    y = y + v sin θ + ω t + 1 αt2 dt (15)  B 0 0 0 2  PLC1 + PLC2 = 2a, where (x, y), (xB, yB) are the coordinates of point PL and point B in the obstacle coordinate system, respectively , and v, θ0 , ω0 and α are the current velocity, heading, angular velocity, and angular acceleration of the USV, respectively. PLC1 + PLC2 is the sum of the distances from PL to the ellipse in the obstacle coordinate system. The a is the semimajor axis of the ellipse. The USV turning right is similar to turning left. The time of starting collision avoidance is tavoid that can be calculated by Equation (16): tavoid = kmax(tL, tR) (16) where k is a coefficient that is greater than 1 to ensure the safety and smooth turning of the USV. When tCPA = tavoid, the USV starts to avoid the obstacle.

3.2.4. Ending Collision Avoidance In this algorithm, the USV needs to travel to the destination along the planned path, therefore, collision avoidance can be ended as long as the USV meets the requirements:

VLOS < VO and Vdest < VO (17)

In Equation (17), VLOS is the velocity vector planned for path following and Vdest is the velocity vector when the USV sails to the target point.

3.2.5. Virtual Obstacle The VO algorithm is set to fully know the obstacle motion information. However, in the actual navigation process, the motion information of the obstacle obtained by the sensor had errors (our USV used LiDAR as the collision avoidance sensor). These errors can cause the USV to collide with the obstacle and, therefore, need to be considered. Appl. Sci. 2020, 10, 2704 10 of 21

Assuming that the velocity vector of the obstacle obtained by the sensor is Vt and the error set of the velocity vector of the obstacle is VE, the actual velocity vector set of the obstacle is:

VP = Vt VE (18) ⊕ where is a Minkowski vector sum operation. Assuming that each element in the VP is a velocity ⊕ vector of a virtual obstacle, and the position and size of the virtual obstacle are the same as the original obstacle, the velocity vector of the virtual obstacle is:

VV = Vt + δV, δV VP (19) ∈ Find the VO set of the obstacle and its associated virtual obstacle and this set is the final VO set of the obstacle. The flowchart of the short-range real-time collision avoidance algorithm is shown in Figure9.

Figure 9. The flowchart of the short-range real-time collision avoidance algorithm. 3.3. LARS Figure 9. The flowchart of the short-range real-time collision avoidance algorithm. The mothership carrying out this mission was equipped with a davit type of LARS. Based on this LARS, we developed a special automatic docking device that solved the problems of docking between our USV and the LARS, thereby improving the accuracy and efficiency of deployment and recovery. Traditional docking devices use a rubber band to eject a rope onto the mothership, and it is necessary to manually reset the rubber band after each ejection, which is cumbersome and dangerous. The use of the rubber band is limited, especially in the environment of high temperature and humidity. Therefore, it has to replace the rubber band frequently, which greatly increases the cost of use. In addition, the elastic force of the rubber band is fixed and the ejection distance is inconvenient to adjust. The newly designed device used an ejection mechanism of the high-pressure gas type instead of the rubber band, which could flexibly change the ejection distance by adjusting the pressure of the compressed air and automatically inflating after each ejection. Compared with the previous ejection mechanism, the new ejection mechanism greatly improved the accuracy, safety, and flexibility of the whole system. The distance of the minimum ejection was 10 m, the distance of the maximum ejection was 100 m, and the inflation time was less than 5 s. The device had both automatic and manual modes. Appl. Sci. 2020, 10, 2704 11 of 21

When in the automatic mode, the USV automatically judged the relative position with the mothership, according to the GPS information, adjusted the ejection parameters, and completed ejection. When in the manual mode, the operator can adjust the ejection parameters of the device through the console on the mothership. In practical applications, the device can safely and quickly deploy or recover 3 tons of USV (7.5 m length) under the sea state 4. As shown in Figure 10, there were three parts in the docking device: (1) The ejection mechanism: The lead and gas cannon were ejected from the USV. (2) The PTZ (Pan/Tilt/Zoom) mechanism: Adjust the launching attitude of the ejection mechanism to ensure that the gas cannon and lead can be launched to the mother ship. The PTZ structure can be adjusted between 0–70◦ and 0–180◦ in the pitch and horizontal directions. (3) The docking mechanism: The interface between the USV and the mothership, which realized the lifting of the USV. As shown in Table2, according to the results of the sea trials, we measured the flying height and time of the gas cannon when the ejection mechanism launched at different initial angles. Appl. Sci. 2020, 10, x FOR PEER REVIEW 11 of 21

(a) (b)

Figure 10.FigureDocking 10. deviceDocking (a) Architecturaldevice (a) Architectural overview of overview the docking of the device. docking (b) Composition device. (b) Composition of the of the docking device. docking device. Table 2. The test results of the ejection mechanism. We focused on the recovery process because the deployment process was simple and warranted θ 30 35 40 45 50 55 60 no further discussion. The ◦recovery◦ process◦ is shown◦ in Figure◦ 11◦ (in the ◦automatic mode). Step (1): Height(m) 7.02 8.72 10.53 12.35 15.03 17.02 19.54 The USV sailsTime to (s)the vicinity0.93 of the0.98 mother1.06 ship,1.15 and the1.26 USV’s 1.42GPS information1.61 is transmitted to the IPC, which calculates the recovery parameters (gas pressure, PTZ mechanism angle, etc.) based on the relative positions of the USV and mothership. Step (2): Start the ejection mechanism and the lead We focused on the recovery process because the deployment process was simple and warranted and gas cannon are ejected to the deck of the mothership. Step (3): Slide the connector on the boom no further discussion. The recovery process is shown in Figure 11 (in the automatic mode). Step (1): of the mothership into the locking device through the lead. Step (4): When the sensor detects the The USV sails to the vicinity of the mother ship, and the USV’s GPS information is transmitted to the connector has fully entered the docking mechanism, the docking mechanism quickly locks the IPC, which calculates the recovery parameters (gas pressure, PTZ mechanism angle, etc.) based on the connector and then lifts the USV. relative positions of the USV and mothership. Step (2): Start the ejection mechanism and the lead and gas cannon are ejected to the deck of the mothership. Step (3): Slide the connector on the boom of the mothership into the locking device through the lead. Step (4):Step2 When the sensor detects the connector has fully enteredStep1 the docking mechanism, the docking mechanism quickly locks the connector and then lifts the USV.

(b) (a)

Step4 Step3

(d)

(c)

Figure 11. Flow chart of the recovery process. (a) Step 1. (b) Step 2. (c) Step 3. (d) Step 4.

Table 2. The test results of the ejection mechanism. Appl. Sci. 2020, 10, x FOR PEER REVIEW 11 of 21

(a) (b)

Figure 10. Docking device (a) Architectural overview of the docking device. (b) Composition of the docking device.

We focused on the recovery process because the deployment process was simple and warranted no further discussion. The recovery process is shown in Figure 11 (in the automatic mode). Step (1): The USV sails to the vicinity of the mother ship, and the USV’s GPS information is transmitted to the IPC, which calculates the recovery parameters (gas pressure, PTZ mechanism angle, etc.) based on the relative positions of the USV and mothership. Step (2): Start the ejection mechanism and the lead and gas cannon are ejected to the deck of the mothership. Step (3): Slide the connector on the boom of the mothership into the locking device through the lead. Step (4): When the sensor detects the Appl.connector Sci. 2020 has, 10, 2704 fully entered the docking mechanism, the docking mechanism quickly locks12 of the 21 connector and then lifts the USV.

Step2 Step1

(b) (a)

Step4 Step3

(d)

(c)

FigureFigure 11.11. FlowFlow chartchart ofof thethe recoveryrecovery process.process. ((aa)) StepStep 1.1. ((bb)) StepStep 2.2. (c) Step 3. ( d) Step 4.

3.4. Sampling System Table 2. The test results of the ejection mechanism. We designed a pump sampling system for our USV, which used a high-precision quantitative peristaltic pump to pump water samples by alternately squeezing and releasing the flexible delivery hose. The delivery hose can be placed at different depths to obtain water samples from different layers. Compared with other types of pumps, the quantitative peristaltic pump did not directly contact the water samples and could precisely control the sampling rate. The maximum sampling rate of the system was 1200 mL/min, and the system could flexibly adjust the sampling rate in the range of 500 mL/min–1000 mL/min. The weight of the system was only 10 kg, the system was 0.3 m 0.2 m 0.5 m (length width height) and was directly installed on the deck of the USV, × × × × which could be removed at any time. In order to improve the reliability of the system, five independent pump sampling systems were installed on the USV. As shown in Figure 12, the system was composed of a peristaltic pump, an automatic retracting hose, and a friction wheel. When the USV received the instruction of the sampling, the friction wheel drove the hose into the water, and then the peristaltic pump alternately squeezed and released the flexible delivery hose to pump the fluid into the collecting box. Finally, when the sampling work was completed, the flexible delivery hose was recovered. Appl. Sci. 2020, 10, x FOR PEER REVIEW 12 of 21

θ 30° 35° 40° 45° 50° 55° 60° Height(m) 7.02 8.72 10.53 12.35 15.03 17.02 19.54 Time (s) 0.93 0.98 1.06 1.15 1.26 1.42 1.61

3.4. Sampling System We designed a pump sampling system for our USV, which used a high-precision quantitative peristaltic pump to pump water samples by alternately squeezing and releasing the flexible delivery hose. The delivery hose can be placed at different depths to obtain water samples from different layers. Compared with other types of pumps, the quantitative peristaltic pump did not directly contact the water samples and could precisely control the sampling rate. The maximum sampling rate of the system was 1200 mL/min, and the system could flexibly adjust the sampling rate in the range of 500 mL/min–1000 mL/min. The weight of the system was only 10 kg, the system was 0.3 m × 0.2 m × 0.5 m (length × width × height) and was directly installed on the deck of the USV, which could be removed at any time. In order to improve the reliability of the system, five independent pump sampling systems were installed on the USV. As shown in Figure 12, the system was composed of a peristaltic pump, an automatic retracting hose, and a friction wheel. When the USV received the instruction of the sampling, the friction wheel drove the hose into the water, and then the peristaltic pump alternately squeezed and released the flexible delivery hose to pump the fluid into the collecting box. Finally, when the sampling work was completed,Appl. Sci. 2020 the, 10 flexible, 2704 delivery hose was recovered. 13 of 21

(b)

(a)

Figure 12. Sampling system (a) Architectural overview of the sampling system. (b) Composition of the Figure 12. Sampling system (a) Architectural overview of the sampling system. (b) Composition of sampling system. the sampling system. 4. Mission 4. Mission 4.1. Mission Plan 4.1. Mission Plan In this mission, our USV was required to cover the mission area with minimum cost and obtain theIn mission this mission, data. In our order USV to was ensure required that the to entire cover shipwreckthe mission could area be with scanned, minimum according cost and to the obtain shape thedata mission of the data. ‘Sanchi’, In order the direction to ensure of thethat wind, the entire and theshipwreck flow of thecould water, be wescanned, determined according a rectangular to the shapemission data area of the of 600‘Sanchi’,600 mthe based direction on the ofGPS the informationwind, and the of theflow sinking of the position.water, we determined a × rectangularAccording mission to area the mission of 600× area,600 m the based distribution on the GPS information information of obstaclesof the sinking (mainly position. the navigation markAccording of the accident to the mission area and area, the rescuethe distribution buoys that information were urgently of obstacles placed after (mainly the accident)the navigation and the markscanning of the widthaccident of the area USV, and the the mission rescue areabuoys was that modeled were urgently by the method placed ofafter the the scan-line accident) fill. Asand shown the scanningin Figure width 13a, afterof the the USV, environment the mission was area modeled, was modeled the mission by the area method was decomposed of the scan-line into afill. series As of Appl. Sci. 2020, 10, x FOR PEER REVIEW 13 of 21 showngrids in with Figure attribute 13a, information,after the environment and the USV was formed modeled, a path the that mission completely area was covered decomposed the environment into a seriesby successively of grids with selecting attribute the information, next scan grid.and the Based USV on formed the rules a path of the that maritime completely measurement, covered the the measurement, the selection principle of the grid node was that of using the method of the global environmentselection principle by successively of the grid selecting node was the that next of using scan grid. the method Based of on the the global rules path of the planning maritime with path planning with artificial initial path constraints. Figure 13b is the planned path of the USV for artificial initial path constraints. Figure 13b is the planned path of the USV for this mission. A total of this mission. A total of 9 lines were planned, each line was 440 m, and the distance between the lines 9 lines were planned, each line was 440 m, and the distance between the lines was 40 m. It is worth was 40 m. It is worth mentioning that when the USV worked according to the planned path, it could mentioning that when the USV worked according to the planned path, it could replan the mission replan the mission area according to the relative position of the shipwreck and the USV after finding area according to the relative position of the shipwreck and the USV after finding the shipwreck, thus the shipwreck, thus improving the work efficiency. improving the work efficiency.

(a) (b)

FigureFigure 13. ((aa)) Environmental Environmental modeling. modeling. ( (bb)) The The planned planned path path of of the the mission. mission.

4.2. Mission Process On the morning of 15 January 2018, we received orders and boarded on the XianYangHong19 (mothership for our USV) to the accident area. After the mission, the Maritime Search and Rescue Center of Shanghai urgently designated an area with a radius of 10 nautical miles around the sinking position of the ‘Sanchi’. All nonrescue vessels were prohibited from entering this area. Figure 14 shows the specific process of the mission. Appl. Sci. 2020, 10, 2704 14 of 21

4.2. Mission Process On the morning of 15 January 2018, we received orders and boarded on the XianYangHong19 (mothership for our USV) to the accident area. After the mission, the Maritime Search and Rescue Center of Shanghai urgently designated an area with a radius of 10 nautical miles around the sinking position of the ‘Sanchi’. All nonrescue vessels were prohibited from entering this area. Figure 14 shows Appl.the specificSci. 2020, process10, x FOR ofPEER the REVIEW mission. 14 of 21

Figure 14. TheThe USV USV in in the the mission: mission: ( (aa)) The The USV USV was was being being deployed deployed in in the the mission mission area. area. ( (bb)) The The USV USV was deployed successfully and sailed to the target. ( c)) The USV was scanning. ( (dd)) Display control interfaceinterface on on the the mothership. mothership. ( (ee)) The The USV USV was was returning returning to to the the mothership. mothership. ( (ff)) The The surface surface of of the the hull hull afterafter the mission was completed.

At 8:20 8:20 on on 18 18 January January 2018, 2018, thethe USVUSV waswas successfullysuccessfully deployed at the edge of of the the mission mission area. area. On the the day day of of the the mission, mission, the wind the wind speed speed was 7.8 was m/ s, 7.8 the m/s, wind the direction wind was direction 32.5◦, the was temperature 32.5°, the temperaturewas 9.3 ◦C, and was the 9.3 wave °C, and height the waswave 1.2 height m. At was 8:30 1.2 the m. USV At8:30 began the to USV work began along to the work planned along paths. the plannedAt 8:57, thepaths. ‘Sanchi’ At 8:57, oil the tanker ‘Sanchi’ was oil found tanker from was the found multibeam from the images. multibeam The imagesimages. showedThe images that showedthe ‘Sanchi’ that oilthe tanker‘Sanchi’ was oil locatedtanker was below located the starboard below the of starboard the USV of and the at USV the and angle at ofthe 10 angle◦ to the of 10°USV’s to the heading. USV’s heading. We narrowed We narrowed the scope the of scope the mission of the mission area based area on based the preliminaryon the preliminary scan results scan resultsand replanned and replanned the path the on path the on basis the of basis the previouslyof the previously planned planned path. Threepath. Three lines were lines addedwere added to the tonorth the north direction, direction, at the at same the same time, time, two north-southtwo north-south direction direction lines lines were were added. added. The The length length of the of theline line was 380was m 380 and m the and spacing the spacing was 270 was m. The 270 USV m. continued The USV to continued work and to then work the and full-precision then the full-precisionmultibeam images multibeam of the ‘Sanchi’ images were of the obtained ‘Sanchi’ at were 9:08. After obtained the shipwreck at 9:08. After was completelythe shipwreck scanned, was completelyfive sampling scanned, points werefive sampling set up in the points horizontal were set and up vertical in the directionhorizontal of theand shipwreck. vertical direction The sampling of the shipwreck.depth was The set to sampling 0 m, 0.3 depth m, 0.5 was m, 0.8set m,to 0 and m, 0.3 1.0 m, m 0.5 and m, the 0.8 interval m, and was1.0 m 10 and m. the At interval 10:58, the was USV 10 m. At 10:58, the USV returned to the mothership and then the technicians immediately extracted the water samples and conducted a comprehensive inspection of the USV.

4.3. Mission Data

4.3.1. Path Following In Figure 15a, the gray straight line indicates the planned path, and the dotted line indicates the actual path of the USV. What we need to explain is that we only focused on the path following error related to the mission, and we did not focus on the path following error during the USV sailing to Appl. Sci. 2020, 10, 2704 15 of 21 returned to the mothership and then the technicians immediately extracted the water samples and conducted a comprehensive inspection of the USV (Supplementary Materials).

4.3. Mission Data

4.3.1. Path Following Appl.In Sci. Figure 2020, 10 15, xa, FOR the PEER gray REVIEW straight line indicates the planned path, and the dotted line indicates 15 theof 21 actual path of the USV. What we need to explain is that we only focused on the path following error relatedthe mission to the area mission, and andreturning we did to not the focus mothership. on the path As shown following in Figure error during 15b, the the maximum USV sailing following to the missionerror was area 3.2150 and returning m. The following to the mothership. error within As shown2 m accounts in Figure for 15 97.70%.b, the maximum The data followingproved that error the wasstraight 3.2150 path m. The following following algorithm error within can ensure 2 m accounts the USV for 97.70%.to have The high data navigation proved that accuracy the straight in this pathmission. following algorithm can ensure the USV to have high navigation accuracy in this mission.

(a) (b)

Figure 15. Path following error of the USV: (a) Path of the USV, (b) changes of the path following error duringFigure the 15. mission. Path following error of the USV: (a) Path of the USV, (b) changes of the path following error during the mission. 4.3.2. Collision Avoidance 4.3.2.Figure Collision 16a–e Avoidance shows a complete collision avoidance process of our USV during the mission. As shownFigure in Figure16a–e 16showsa, the a USV complete was sailing collision normally avoidance according process to the of plannedour USV path, during and the a large mission. rescue As vesselshown was in Figure coming 16a, from the the USV right was side sailing of the normally USV. As according shown in Figureto the planned 15b–d, when path, theand USV a large detected rescue thevessel vessel, was the coming collision from avoidance the right process side started, of the theUSV. USV As began shown to inchange Figure the 15b–d, path according when the to USV the collisiondetected avoidance the vessel, parameters, the collision steered avoidance to the right, process and detoured started, the from USV the rear began side to of change the vessel the until path itaccording successfully to andthe collision safely bypassed avoidance the parameters, vessel, and thensteered returned to the to right, the plannedand detoured path. Figurefrom the 16 erear shows side theof changethe vessel of theuntil path it successfully during the whole and safely collision bypassed avoidance the vessel, process. and then returned to the planned path.Then, Figure we 16e analyzed shows the change changes of of the the path USV’s during heading the whole angle collision and lateral avoidance offset during process. collision avoidance process. As can be seen from Figure 17a,b, during the initial stage, the USV was sailing along the path with an angular direction of 70.219◦, while the lateral offset was controlled within 3 m. Starting from 55 s, the USV detected the obstacle and then turned to the right. At this time, the lateral offset started to decrease (the lateral offset was positive on the left side of the path), and, at most, the lateral offset was about 20 m. After completely avoiding the obstacle, the USV began to quickly − adjust its path, turn left, and return to the planned path. So the lateral offset gradually increased until it returned to near 0, indicating that the USV returned to the planned path. As shown in Figure 17c, during the whole collision avoidance process, the nearest distance between the USV and the obstacle was 16 m, which is sufficient to ensure navigation safety.

Figure 16. Process of the collision avoidance: (a) Start collision avoidance, (b,c) collision avoidance, (d) return to the test path after avoiding the obstacle, (e) path in the process of collision avoidance.

Then, we analyzed the changes of the USV’s heading angle and lateral offset during collision avoidance process. As can be seen from Figure 17a,b, during the initial stage, the USV was sailing Appl. Sci. 2020, 10, x FOR PEER REVIEW 15 of 21 the mission area and returning to the mothership. As shown in Figure 15b, the maximum following error was 3.2150 m. The following error within 2 m accounts for 97.70%. The data proved that the straight path following algorithm can ensure the USV to have high navigation accuracy in this mission.

(a) (b)

Figure 15. Path following error of the USV: (a) Path of the USV, (b) changes of the path following error during the mission.

4.3.2. Collision Avoidance Figure 16a–e shows a complete collision avoidance process of our USV during the mission. As shown in Figure 16a, the USV was sailing normally according to the planned path, and a large rescue vessel was coming from the right side of the USV. As shown in Figure 15b–d, when the USV detected the vessel, the collision avoidance process started, the USV began to change the path according to the collision avoidance parameters, steered to the right, and detoured from the rear side of the vessel until it successfully and safely bypassed the vessel, and then returned to the planned Appl. Sci. 2020 10 path. Figure ,16e, 2704 shows the change of the path during the whole collision avoidance process. 16 of 21

Appl. Sci. 2020, 10, x FOR PEER REVIEW 16 of 21

along the path with an angular direction of 70.219°, while the lateral offset was controlled within 3 m. Starting from 55 s, the USV detected the obstacle and then turned to the right. At this time, the lateral offset started to decrease (the lateral offset was positive on the left side of the path), and, at most, the lateral offset was about −20 m. After completely avoiding the obstacle, the USV began to quickly adjust its path, turn left, and return to the planned path. So the lateral offset gradually increased until it returned to near 0, indicating that the USV returned to the planned path. As shown Figure 16. Process of the collision avoidance: (a) Start collision avoidance, (b,c) collision avoidance, in FigureFigure 17c,16. Process during of the the whole collision collision avoidance: avoidance (a) Start process,collision avoidance,the nearest (b ,distancec) collision between avoidance, the USV (d) return to the test path after avoiding the obstacle, (e) path in the process of collision avoidance. and( dthe) return obstacle to the was test 16 path m, after which avoiding is sufficient the obstacle, to ensure (e) path navigation in the process safety. of collision avoidance.

Then, we analyzed the changes of the USV’s heading angle and lateral offset during collision avoidance process. As can be seen from Figure 17a,b, during the initial stage, the USV was sailing

(a) (b)

(c)

FigureFigure 17. 17.Collision Collision avoidance avoidance results results of theof the USV. USV. (a) Changes (a) Changes in heading in heading angle. angle. (b) Changes (b) Changes in lateral in olateralffset.( offset.c) Change (c) Change in distance in distance between between the USV the and USV the and obstacle the obstacle center. center.

4.3.3.4.3.3. ScanScan FigureFigure 1818aa represents the front front view view of of the the shipwreck, shipwreck, where where you you can can see see the the total total length length of the of theshipwreck shipwreck was was 274 274 m. m. Figure Figure 18b 18 showsb shows the the side side view view of of the the shipwreck. shipwreck. We We can can see that that the the prominentprominent cockpitcockpit waswas perpendicularperpendicular toto thethe surfacesurface ofof thethe seabed,seabed, indicatingindicating thatthat thethe shipwreckshipwreck waswas alsoalso perpendicularperpendicular toto thethe surfacesurface ofof thethe seabed.seabed. Appl. Sci. 2020, 10, 2704 17 of 21 Appl. Sci. 2020, 10, x FOR PEER REVIEW 17 of 21

(a)

(b)

FigureFigure 18. 18. ‘Sanchi’‘Sanchi’ multibeam multibeam view: view: (a (a) )Shipwreck Shipwreck front front view, view, ( (bb)) shipwreck shipwreck side side view. view.

ByBy preliminary preliminary judgment, judgment, the the oil oil spill spill point point was was located located in in the the position position where where the the hull hull was was hit hit in in thethe accident. accident. It It is is well well known known that that a a huge huge impact impact can can cause cause a a partial partial structural structural loss loss of of the the hull, hull, so so the the locationlocation where where the structural lossloss occursoccurs isis thethe pointpoint ofof thethe impact.impact. ReferringReferring to to Figure Figure 19 19a,a, we we found found a agap gap in in the the front front view. view. In In order order to furtherto further judge judge whether whether the the gap gap was was caused caused by the by collision,the collision, we took we tookthe cross-section the cross-section 1 at the1 at gap the and gap took and anothertook another cross-section cross-section 2 near 2 the near 1, thenthe 1, generated then generated the height the heightchange change diagram diagram of the two of cross-sections.the two cross-sections. Referring Referring to the Figure to the19b,c, Figure we found 19b,c, that we thefound cross-section that the cross-section1 had a significant 1 had dropa significant in height drop at the in height same location at the same compared location with compared the cross-section with the cross-section 2, indicating 2,that indicating the cross-section that the cross-section 1 had a structural 1 had loss.a structural Combined loss. withCombined the dimensional with the dimensional structure data structure of the data‘Sanchi’ of the oil tanker,‘Sanchi’ we oil confirmed tanker, we that confirmed the gap inthat Figure the gap19a wasin Figure caused 19a by was the collision,caused by and the the collision, oil spill andpoint the was oil atspill this point gap. was at this gap. TheThe multibeam multibeam images images not not only only helped helped us us confirm confirm the the position position and and posture posture of of the the shipwreck, shipwreck, butbut also also found found the the position position of of the the impact impact to to confirm confirm the the point point of of the oil spill. On On the the one one hand, hand, this this informationinformation helped helped us us infer infer the the oil oil spill spill rate rate of ofthe the shipwreck shipwreck and and adjust adjust the themission mission plan plan in time. in time. On theOn other the other hand, hand, it provided it provided strong strong evidence evidence for future for future accident accident investigation. investigation. Appl.Appl. Sci.Sci.2020 2020,,10 10,, 2704 x FOR PEER REVIEW 18 18of of21 21

(a)

(b)

(c)

FigureFigure 19.19.Cross-sectional Cross-sectional view view of of ‘Sanchi’ ‘Sanchi’ multibeam: multibeam: (a )(a Cross-section) Cross-section position, position, (b )( heightb) height change change of cross-sectionsof cross-sections 1, ( c1,) height(c) height change change of cross-sections of cross-sections 2. 2.

4.3.4.4.3.4. WaterWater SamplingSampling OurOur USVUSV collectedcollected aa totaltotal ofof 55 bottlesbottles ofof thethe waterwater samplessamples (2000(2000 mLmL perper bottle)bottle) atat 55 didifferentfferent locations.locations. After After the the USV USV returned returned to the to mothership, the mothership, the China the State China Oceanic State Administration Oceanic Administration analyzed theanalyzed water samples.the water Figuresamples. 20 Figureshows 20 the shows collected the watercollected samples. water samples. Since oil Since can change oil can and change destroy and thedestroy normal the ecological normal ecological balance in balance the water, in the serious water, oil serious spill accidents oil spill often accidents destroy often the destroy ecological the environmentecological environment for decades. Therefore, for decades. water Therefore, quality monitoring water quality must bemonitoring carried out must through be thecarried mission out andthrough operate the mission for a long and time operate after for this a long mission. time Evenafter this if the mission. oil concentration Even if the oil in theconcentration water basically in the returnswater basically to normal, returns monitoring to normal, cannot monitoring be stopped cannot until be the stopped ecological until environment the ecological is fully environment restored. is fully restored. Appl. Sci. 2020, 10, 2704 19 of 21 Appl. Sci. 2020, 10, x FOR PEER REVIEW 19 of 21

Figure 20. The water samples. Figure 20. The water samples. 5. Conclusions 5. Conclusion This paper described the application of our USV in the emergency response mission for the ‘Sanchi’This oil paper tanker described collision andthe explosionapplication accident. of our USV The improvedin the emergency algorithms response of navigation mission control, for the an‘Sanchi’ improved oil tanker automatic collision LARS, and and explosion a new specialaccident. sampling The improved system algorithms allowed the of USVnavigation to be quicklycontrol, deployedan improved to the automatic mission area, LARS, and and enabled a new the special USV to sampling have higher system navigation allowed accuracy the USV and to navigationbe quickly safetydeployed in the to mission. the mission In this area, mission, and the enabled USV performed the USV the to real-time have higher scanning navigation and sampling, accuracy which and providednavigation important safety in information the mission. for In determining this mission, the the exact USV location performed of the the shipwreck, real-time locating scanning the and oil spillsampling, point, estimating which provided the amount important of oil spilled, information salvaging for the determining shipwreck, and the evaluating exact location the pollution. of the Theshipwreck, information locating not onlythe oil provided spill point, a lot estimating of help to the rescuers, amount but of alsooil spilled, offered salvaging a scientific the basis shipwreck, for the follow-upand evaluating work (suchthe pollution. as ecological The recovery,information accident not only identification, provided a etc.). lot of help to rescuers, but also offeredAlthough a scientific the USV basis completed for the follow-up the mission, work we (such also foundas ecological some new recovery, issues accident in the mission, identification, such as theetc.). stability of the USV was insufficient. Especially in poor sea states, the negative impact of the environmentalAlthough disturbances the USV completed (winds, waves,the mission, and currents, we also etc.)found on some the stability new issues was particularlyin the mission, evident. such Theas the capability stability of of a singlethe USV USV was was insufficient. also limited. Especially Due to the in limitationspoor sea states, of the the speed negative and the impact payload, of the a singleenvironmental USV had to disturbances take a long time (winds, to complete waves, aand complex currents, mission. etc.) Once on the the stability USV breaks was down particularly in the processevident. of The working, capability the missionof a single must USV be was suspended. also limited. Due to the limitations of the speed and the payload,We will a single continue USV to had solve to thetake above a long issues time byto activelycomplete predicting a complex the mission. environmental Once the disturbances’ USV breaks informationdown in the andprocess establishing of working, a multiplethe mission USVs’ must system. be suspended. Active prediction of the environmental disturbanceWe will information continue for to the solve USVs the can calculate above issues the compensation by actively instructions predicting in the advance environmental according todisturbances’ environmental information disturbances, and so establishing that the stability a multiple of USVs USVs’ can be improved.system. Active Multiple prediction USVs’ system of the withenvironmental the cooperation disturbance between information the USVs canfor the enhance USVs thecan USVs’calculate robustness the compensation and reliability, instructions improve in missionadvance performance, according to reduce environmental costs, and disturbances, provide the best so that mission the policy.stability of USVs can be improved. Multiple USVs’ system with the cooperation between the USVs can enhance the USVs’ robustness Supplementaryand reliability, Materials: improve Themission following performance, are available reduce online costs, at http: and//www.mdpi.com provide the best/2076-3417 mission/10 policy./8/2704 /s1, Video: Collision avoidance, Deployment, Recovery. Author Contributions: Author Contributions: H.P., Y.L., J.L., S.X., Y.P. designed USVs; Y.Y. (Yang Yang), X.L., C.Z., J.K., J.C. and D.Q. carriedHuayan out Pu, the Yuan missions; Liu, Jun Y.L., Luo, Z.S., Shaorong W.S. analyzed Xie, Yan mission Peng results. * designed H.P., Y.Y.USVs; (Yi Yang Yang), Yang, S.G. analyzedXiaomao sequencingLi, Chuang data and developed analysis tools. Y.L. wrote the manuscript. All authors have read and agreed to the published Zhu, Jun Ke, Jianxiang Cui and Dong Qu carried out the missions; Yuan Liu, Zhou Su, Wenyun Shao analyzed version of the manuscript. mission results. Huayan Pu, Yi Yang, Shouwei Gao analyzed sequencing data and developed analysis tools. Funding: This work was supported in part by the National Natural Science Foundation of China (9174810194). Yuan Liu wrote the manuscript. Acknowledgments: The authors wish to thank the State Oceanic Administration of China, the Maritime Search andFunding: Rescue This Center work of was China, supported and the in Maritimepart by the Search National and Natural Rescue Science Center Foundation of Shanghai of thatChina have (9174810194). helped in the missions. Acknowledgments: Conflicts of Interest: The authors declare there is no conflicts of interest regarding the publication of this paper. The authors wish to thank the State Oceanic Administration of China, the Maritime Search and Rescue Center of China, and the Maritime Search and Rescue Center of Shanghai that have helped in the missions. Appl. Sci. 2020, 10, 2704 20 of 21

Abbreviations

Unmanned Surface Vehicle USV launch and recovery system LARS Autonomous Underwater Vehicle AUV Remote Operated Vehicle ROV Knot kn Global Positioning System GPS Inertial Navigation System INS Light Detection and Ranging LiDAR

References

1. Tiron, R. High-Speed Unmanned Craft Eyed for Surveillance Role. Natl. Def. 2002, 5, 27. 2. Accurate Automation Corporation. Available online: https://www.accurateautomation.com/content/Home (accessed on 1 December 2019). 3. C-Target3. Available online: https://www.asvglobal.com/product/c-target-3/ (accessed on 5 December 2019). 4. Caccia, M.; Bibuli, M.; Bono, R.; Bruzzone, G. Basic navigation, guidance and control of an Unmanned Surface Vehicle. Auton. Robot 2008, 25, 349–365. [CrossRef] 5. Furfaro, T.C.; Dusek, J.E.; Von Ellenrieder, K.D. Design, Construction, and Initial Testing of an Autonomous Surface Vehicle for Riverine and Coastal Reconnaissance. In Proceedings of the MTS/IEEE International Conference on OCEANS, Biloxi–Marine Technology for the Future: Global and local Challenges, Piscataway, NJ, USA, 26–29 October 2009; pp. 1–6. 6. Anon. 2014. Available online: http://www.asvglobal.com/ (accessed on 21 August 2015). 7. Naeem, W.; Xu, T.; Sutton, R.; Tiano, A. The design of a navigation, guidance, and control system for an unmanned surface vehicle for environmental monitoring. Proc. Inst. Mech. Eng. Part M J. Eng. Marit. Environ. 2008, 222, 67–79. [CrossRef] 8. Caccia, M.; Bibuli, M.; Bono, R.; Bruzzone, G.; Bruzzone, G.; Spirandelli, E. Unmanned surface vehicle for coastal and protected waters applications: The charlie project. Mar. Technol. Soc. J. 2007, 41, 62–71. [CrossRef] 9. Caccia, M. Modelling and identification of the Charlie2005 ASC. In Proceedings of the 14th Mediterranean Conference on Control and Automation, Ancona, Italy, 28–30 June 2006; pp. 1–6. 10. Roberts, G.N.; Sutton, R. (Eds.) Advances in Unmanned Marine Vehicles; The Institution of Engineering and Technology: London, UK, 2006. 11. Gomes, P.; Silvestre, C.; Pascoal, A.; Cunha, R. A Path-Following Controller for the DELFIMx Autonomous Surface Craft. In Proceedings of the 7th IFAC Manoeuvring and Control of Marine Craft, Lisbon, Portugal, 20–22 September 2006. 12. Motwani, A. A Survey of Uninhabited Surface Vehicles; MIDAS Technical Report. MIDAS.SMSE.2012.TR.001; Plymouth University: Plymouth, UK, 2012. 13. Murray, J. Sentry—An Unmanned Swimmer Intercept System. DTIC Document; QinetiQ North America Inc., Woburn Ma Technology Solutions Group: Woburn, MA, USA, 2008. 14. Sonnenburg, C.R. Modeling, Identification, and Control of an Unmanned Surface Vehicle. Ph.D. Thesis, Virginia Polytechnic Institute and State University, Blacksburg, VA, USA, 2012. 15. Yang, W.R.; Chen, C.Y.; Hsu, C.M.; Tseng, C.J.; Yang, W.C. Multifunctional inshore survey platform with unmanned surface vehicles. Int. J. Autom. Smart Technol. 2011, 1, 19–25. [CrossRef] 16. Vasilijevic, A.; Stilinovic, N.; Nad, D.; Mandic, F.; Miskovic, N.; Vukic, Z. AUV Based Mobile Fluorometers: System for Underwater Oil-Spill Detection and Quantification. In Proceedings of the Sensors Applications Symposium, Zadar, Croatia, 13–15 April 2015; pp. 1–6. 17. Valentine, M.M.; Benfield, M.C. Characterization of epibenthic and demersal megafauna at Mississippi Canyon 252 shortly after the Deepwater Horizon Oil Spill. Mar. Pollut. Bull. 2013, 77, 196–209. [CrossRef] [PubMed] 18. Murphy, R.R.; Steimle, E.; Griffin, C.; Cullins, C.; Hall, M.; Pratt, K. Cooperative use of unmanned sea surface and micro aerial vehicles at Hurricane Wilma. J. Field Robot. 2008, 25, 164–180. [CrossRef] 19. Park, W.G.; Jang, J.H.; Chun, H.H.; Kim, M.C. Numerical flow and performance analysis of waterjet flow and performance analysis of water-jet propulsion system. Ocean Eng. 2005, 32, 1740–1761. [CrossRef] Appl. Sci. 2020, 10, 2704 21 of 21

20. Kim, M.C.; Chun, H.H.; Kim, H.Y.; Park, W.K.; Jung, U.H. Comparison of waterjet performance in tracked vehicles by impeller diameter. Ocean Eng. 2009, 36, 1438–1445. [CrossRef] 21. Fox, D.; Burgard, W.; Thrun, S. The dynamic window approach to collision avoidance. IEEE Robot. Autom. Mag. 1997, 4, 23–33. [CrossRef] 22. Fiorini, P. Motion planning in dynamic environments using velocity obstacles. Int. J. Robot. Res. 1998, 17, 760–772. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).