GENERAL POSS: TO SWARM, BEYOND AG/MINING P.10 OR NOT TO SWARM P.16 TAKES OFF P.58 USE CASES

inside June/July 2021 unmanned systems INSIDE ENGINEERING, POLICY AND PRACTICE InsideUnmannedSystems.com

SPECIAL COLLABORATION ISSUE JPL AND AEROVIRONMENT

MARSon

SUCCEED AT SMART COLLABORATION

PUBLISHED BY AUTONOMOUS MEDIA, LLC COLLABORATION Mars Helicopter Project

INSIDE THE INGENUITY HELICOPTER: Teamwork on Mars pril 19th saw what some have punching above your weight. The Mars Helicopter christened “a second Wright “Now that Ingenuity is actually flying at Project, a.k.a. A Brothers moment”—namely, the Mars, we can begin to assess how things Ingenuity, lifts off successful first powered controlled flight stack up against expectations,” noted Håvard from the Martian by an aircraft on another world. Reaching Grip, Mars helicopter chief pilot for NASA’s surface, near the Perseverance rover. Mars on the underside of the Perseverance Jet Propulsion Laboratory (JPL). rover, the tiny, autonomous Mars Ingenuity Ingenuity represents the years-long Helicopter (5.4" x 7.7" x 6.4") spun its 4-foot collaboration between NASA/JPL, ma- rotors and hovered 10 feet off the ground jor unmanned systems manufacturer for 30 seconds. By its third flight, a few AeroVironment and a bevy of other compa- days later, Ingenuity would rise 16 feet (5 nies. The articles that follow chronicle how meters) up, and fly 164 feet (50 meters) at JPL created the craft’s unique navigation sys- a top speed of 6.6 ft/sec (2 m/sec). Back in tem and how AeroVironment’s engineering 1903, the Wright Brothers logged 120 feet team stepped up in the electrical, mechani- to complete the first controlled heavier- cal, systems and vehicle flight control areas. than-air powered flight. Now, squaring “Working with JPL, I think we’ve that circle, Ingenuity carries a piece of fab- learned a lot,” AeroVironment engineer- ric from the Wright Flyer’s wing, and its ing lead Ben Pipenberg said. “It’s a very flight site is called Wright Brothers Field. AV project, it’s this weird, first-of-its-kind Six weeks and six flights into its mission vehicle, but JPL is the best at what they do as we write, Ingenuity has demonstrated by far, in terms of space robotics, planetary the ability to fly on a planet more than 170 robotics, and there was a huge amount we million miles from earth in an atmosphere were able to learn from them that we’ll be 1% as thick as ours. The near-miniature ve- able to pull into a lot of our work in the fu- hicle has proved to be an intrepid explorer ture. It was really good teaming. even as it’s survived a computer anomaly “There’s a huge amount of pride, a lot of on its most recent mission. Talk about excitement, that it’s flying on Mars.” 

Photo courtesy of NASA/JPL.  www.insideunmannedsystems.com  June/July 2021 COLLABORATION Mars Helicopter Project Mars Helicopter Project COLLABORATION

flights began and ended within an area that had been pre-inspected and deter- mined to be safe in terms of obstacles and ground slope. Ingenuity conducted five flights accord- ing to its programmed lifeline across a pe- riod of 31 Earth days, or 30 sols on Mars. “Now that Ingenuity Then came the surprise ending-to-date, but more on that later. is actually flying For the helicopter’s pre-arranged auton- omous test flights, under the NASA rubric at Mars, we can begin to assess of “technology demonstration,” it took off, how things stack up climbed, hovered, translated between a set against expectations.” of waypoints, then descended to land again (see Figure 1). Although the helicopter did Håvard Grip, Mars helicopter chief pilot, JPL operate independently during flight, the waypoints were specified from Earth prior to flight. FIGURE 1 Illustration of a Mars Helicopter flight, beginning and ending in the same AUTONOMY? pre-inspected safe area. This, however, raises an interesting and somewhat subtle point: is Ingenuity truly Pipenberg, AeroVironment’s engineer- autonomous? ing lead on the Ingenuity project. “When It depends on your definition. Engineers Perseverance landed, it used terrain-rel- at AeroVironment, which constructed ma- ative navigation, and it was making de- jor elements of the helicopter but was not cisions based on outside observable data involved in the guidance, navigation and that it was collecting without human When Ingenuity, the little Autonomously, in other words. Radio sig- control (GNC) system design, weighed in input. That would be an autonomous helicopter that could, sprang Autonomously Alone nals from NASA Command take 15 min- on the issue. system. Ingenuity is not doing that. It’s from the Martian surface JPL: on the Red Planet utes and 27 seconds to travel the 173 million “It certainly is making autonomous deci- essentially using VIO—visual-inertial into the wispy thin Martian miles (278.4 million kilometers) to Mars. sions [in managing rotor speed and pitch] odometry—just to navigate over the o make this happen, NASA invested Mode Commander, setting the overall Once on the surface, the more well-en- to get more cyclic to overcome a wind gust,” ground in a pre-determined flight path, atmosphere, it knocked down $85 million to build Ingenuity, ac- mode for the flight control system; the dowed Perseverance rover served as a com- said Jeremy Tyler, senior aeromechanical uploaded from Earth.” all kinds of firsts. The first T commodate it onboard Perseverance Guidance subsystem, providing refer- munications relay link so the helicopter and engineer. “It’s managing its altitude, it’s Tyler concurred, after a fashion. “It’s powered, controlled flight for the long interplanetary flight and para- ence trajectories for the flight path; the Mission Team on Earth could communicate. managing its position, all by itself without doing its own simple autonomy. But cer- on another planet. The first chute deployment, and operate it once it Navigation subsystem, giving estimates It passed flight instructions from NASA’s any external intervention.” tainly no sophisticated mission planning autonomous flight. The first reached distant Mars. of the vehicle state; and the Control sub- Jet Propulsion Laboratory in Pasadena, “It’s inherently unstable,” added Matt or decision-making.” use of an inertial navigation There’s plenty to marvel at in this un- system, commanding the actuators based California, to Ingenuity. From a Martian Keennon, technical lead for rotor system dertaking, which took the fertile minds of on the reference trajectories and the ve- hillock 65 meters away, the four-wheeled development. “It can’t fly for a half-second THE NAVIGATION SYSTEM system and visual odometry NASA’s Jet Propulsion Laboratory (JPL), hicle state. rover observed and recorded its four-bladed without making decisions based on the in- Engineers at JPL under the direction of across an alien world. NASA Ames Research Center, NASA The specific challenges for the naviga- offspring’s history-making flights. ertial measurement unit [IMU] and driv- Håvard Grip, Mars helicopter chief pi- Langley Research Center and companies tion system onboard the UAV include: While hovering on its four initial flights, ing the control system.” lot, developed and assembled the visual- that included AeroVironment, Inc. (see • A lack of global navigation aids, such as the helicopter’s navigation camera and “There’s no [navigation] decisions inertial navigation system emphasizing by Alan Cameron, PNT Editor accompanying feature) on a six-year jour- GPS or a strong magnetic field. laser altimeter fed information into the being made onboard,” countered Ben robustness, but with a correspondingly ney from inspiration to realization. Awe • A large communication time lag navigation computer to ensure Ingenuity will be confined within this article to the between Earth and Mars, preventing remained not only level, but within the phenomenon of Ingenuity’s navigation real-time communication during flight. borders of its 10x10 meter airfield—a (TOP RIGHT) The Ingenuity Mars Helicopter’s navigation camera system. • A harsh radiation environment patch of extraterrestrial real estate cho- “As we continue with our flights captures the helicopter’s shadow on that can adversely affect computing sen for its flatness and lack of obstruc- on Mars, we will keep digging deeper into the data…” the surface of Jezero Crater during the NAVIGATING THE SUBSYSTEMS elements. tions. Because landing hazard avoidance Håvard Grip, Mars helicopter chief pilot, JPL rotorcraft’s second experimental test The Mars copter’s flight control system Because of the time-lag challenge, was not prioritized for this technology flight on April 22. consists of four main subsystems: the Ingenuity has to perform on its own. demonstration, each of those four initial

Photos and figures courtesy of NASA/JPL-Caltech.  www.insideunmannedsystems.com  June/July 2021 June/July 2021  www.insideunmannedsystems.com  COLLABORATION Mars Helicopter Project Mars Helicopter Project COLLABORATION

limited position accuracy and ability to facing view for terrain images, not used for its own and relied on a set of flight-con- tion camera tracks visual features on the navigate in complex terrain. In particu- navigation. trol algorithms that we developed here on ground, under the assumption that all fea- lar, the system assumes that features ob- Figure 2 shows Ingenuity’s avionics Earth, before Ingenuity was even launched tures are located on a ground plane with a served by the navigation camera lie on an system architecture. A radiation-tolerant to Mars.” known slope. This provides horizontal ve- approximate ground plane with known field-programmable gate array (FPGA) When the copter rests on the ground, locity as well as roll and pitch attitude, and slope. This is why the landing field was function routes sensor data and traffic preparing to take off, the inclinometer helps limit the drift in horizontal position chosen and why the first four flights did between other computing elements and estimates initial roll and pitch attitude. and yaw angle. not venture beyond its bounds. The flights performs low-level actuator control. Most Based on this, initial estimates of the However, the latter two measurements took place over relatively flat terrain, with of the flight control software is hosted on accelerometer and gyro biases are also have no absolute reference, and their es- short-term height variations on the order the flight computer (FC). obtained. timates are subject to long-term drift. of 10% of the flight height. A separate navigation computer Once the vehicle is in motion, integra- Therefore, shortly before touchdown at (NC), a 2.26 GHz quad-core Qualcomm tion of the IMU measurements is used to the end of each flight, a navigation cam- The navigation sensors Ingenuity Snapdragon 801 processor, provides the estimate changes in position, velocity and era image is stored for later transmission carries are: throughput for vision-based navigation. attitude. Only the IMU is used for this on Earth, so that an absolute position and • Bosch Sensortech BMI160 IMU, for On the NC, one core is devoted to camera- critical second, measuring acceleration heading fix can be obtained by comparison measuring 3-axis accelerations at 1600 image processing and another to the navi- and angular rates. After the helicopter to the known terrain. Hz and angular rates at 3200 Hz. gation filter, while the remaining cores are reaches 1 meter off the ground, the laser “To develop the flight control algorithms,” • Garmin Lidar-Lite-V3 laser rangefinder used for other activity. rangefinder and downward-looking cam- Grip wrote in a NASA blog post updating (LRF), for measuring distance to the The visual-inertial navigation system era are added to the navigation solution. Ingenuity’s fans, “we performed detailed ground at 50 Hz. provides the control system with real-time This precaution springs from pre-mission modeling and computer simulation in or- • Downward-looking 640 x 480 grayscale estimates of the vehicle state: position, concern that the LRF and camera might der to understand how a helicopter would camera with an Omnivision OV7251 velocity, attitude and angular rates. The be obscured by dust kicked up by the cop- behave in a Martian environment. We fol- FIGURE 3 Illustration of the flight control software implementation on the flight global-shutter sensor, providing images state estimate is based on fusing informa- ter blades. The IMU will not output great lowed that up with testing in a massive avionics, with the flow of sensor, actuator and state estimate information. Non-real-time at 30 Hz. tion from the onboard IMU, inclinometer, accuracy in the long-run, but because 25-meter-tall, 7.5-meter-diameter vacuum components are indicated with dashed lines. • MuRata SCA100T-D02 inclinometer, for LRF and navigation camera. Ingenuity takes only a couple of seconds chamber here at JPL, where we replicate measuring roll and pitch attitude prior to reach 1 meter, “we can make it work,” the Martian atmosphere. But in all of that reset as features are lost, MAVeN is effec- phase, this is an acceptable tradeoff, be- to flight. HEAD TO THE CHOPPER Grip said. Ingenuity then starts using its work, we could only approximate certain tively a long-baseline visual odometry al- cause accuracy degradation is graceful All are commercial off-the-shelf (COTS) “Before each of Ingenuity’s test flights,” full suite of sensors. aspects of the environment. Now that gorithm: the relative position and attitude and the algorithm has proven to be highly miniature sensors, largely developed for the Grip told Inside Unmanned Systems, “we During hover flight, Ingenuity on its Ingenuity is actually flying at Mars, we can between the two images are measured, robust in both simulation and experiments. cell phone and lightweight drone markets. uploaded instructions describing precisely semi-autonomous own attempts to main- begin to assess how things stack up against but not the absolute position and attitude. Feature detection in base images is Ingenuity also carried a second camera, what the flight should look like. But when tain a constant altitude, heading and po- expectations.” Absolute position and attitude error, in this performed with an implementation of the a 13-megapixel color camera with horizon- it came time to fly, the helicopter was on sition. The JPL team has to rely on the The MAVeN navigation algorithm used case horizontal position and yaw, grow FAST algorithm [30], which selects corner- copter’s estimates on how well it performs “has no absolute references to any land- over time. The LRF provides vertical posi- like features that have sufficient contrast this task, as there is limited to no basis for marks,” according to Grip. “It always op- tion, which bounds vertical position error. between a center pixel and a contiguous ground truth. But the available data shows erates against a base frame where it sees In addition, the visual features and flat- arc surrounding the center pixel. An al- that Ingenuity holds its altitude extremely a bunch of features and tracks them over plane assumption provide observability of gorithm estimates the displacement of a well in hover, to within approximately 1 a limited set of search frames. When it’s absolute pitch and roll when the vehicle is template from one image to the next, us- centimeter, and its heading to within less done, it requires a completely new base moving. ing a gradient-based search algorithm that than 1.5 degrees. Horizontal position can frame. It is always tracking in a relative A key advantage of MAVeN over other minimizes the difference in pixel intensity vary up to approximately 25 centimeters, sense, never tied back to a global frame. simultaneous localization and mapping (see Figure 3). which the team attributes to wind gusts MAVeN is implemented as an Extended (SLAM) algorithms is that the state only on the Red Planet. Kalman Filter (EKF) that also uses the needs to be augmented with six scalar ele- BRINGING IT ALL BACK HOME difference between the predicted and ments—three for position and three for at- Landing is an altogether delicate matter. CRUISE CONTROL measured LRF range. MAVeN has a state titude. The LRF and an assumed ground A rapid sequence of events takes place Because of the relatively low accuracy of vector with seven components: position, plane enable MAVeN to estimate 3D po- as Ingenuity descends toward the ground. MEMS-based IMUs, navigation aids must velocity, attitude, IMU accelerometer bias, sition and velocity without introducing a “First, a steady descent rate of 1 meter per bound the growth in navigation errors as IMU gyro bias, base image position and scale ambiguity. second is established,” Grip wrote. “Once the copter cruises. The LRF provides range base image attitude, for a total of 21 scalar The two main disadvantages of MAVeN the vehicle estimates that the legs are measurements between the vehicle and components. are sensitivity to rough terrain, due to within 1 meter of the ground, the algo- the terrain below, giving vertical velocity MAVeN only tracks features between the ground-plane assumption, and long- rithms stop using the navigation camera and position. With the aid of the MaVeN the current search image and the base im- term drift in position and heading. For and altimeter for estimation, relying on FIGURE 2 The Mars Helicopter avionics architecture. feature-tracking algorithm, the naviga- age. Because the base frame is frequently Ingenuity’s technology demonstration the IMU in the same way as on takeoff. As

 www.insideunmannedsystems.com  June/July 2021 June/July 2021  www.insideunmannedsystems.com  COLLABORATION Mars Helicopter Project Mars Helicopter Project COLLABORATION

with takeoff, this avoids dust obscuration, an end, its duty done. Its parent and ride role,” Grip said, “doing flights every two to but it also serves another purpose: by rely- to Mars, the four-wheeled Perseverance three weeks, to demo operational capabil- ing only on the IMU, we expect to have a rover, would continue for two more years ity, at higher risk, and focused more on very smooth and continuous estimate of to explore the Jezero Crater, site of a lake aerial imaging capabilities. our vertical velocity, which is important 3.9 billion years ago, seeking traces of an- “These flights are stretching Ingenuity’s in order to avoid detecting touchdown cient microbial life. Ingenuity would perch capability in terms of altitude, distance prematurely. motionless forever upon the Martian land- and speed. We’ve covered our basics, “About half a second after the switch to scape, the lonely one. shown that a helicopter can fly on Mars, IMU-only, when the legs are estimated to But wait. nicely and confidently. We’re now stretch- be within 0.5 meters of the ground, the “On the last flight, we actually flew some- ing the parameters of those flights with the touchdown detection is armed. Ingenuity where else,” Grip said. “We had scouted hardware and software that we have on the will now consider touchdown to have that terrain previously with the helicopter. helicopter.” occurred as soon as the descent veloc- “In that scouting flight, No. 4, we took The increased speed over ground af- ity drops by 25 centimeters per second or images using the high-resolution return- fects the navigation system and how the more. Once Ingenuity meets the ground, to-Earth color camera. We could see on our features the camera is tracking move that drop in descent velocity happens target airfield, individual rocks, ripples, fea- through the field of view. Additionally, rapidly. At that point, the flight control tures, that we then georeferenced against a new flights will break the parameter of system stops trying to control the motion low-resolution satellite image, so we knew flying over relatively flat terrain. “We of the helicopter and commands the col- exactly where those features were in a may fly over less flat terrain, that will lective control to the lowest possible blade global frame. When we went back on flight challenge the navigation algorithm. How pitch to produce close to zero thrust. The 5, we could use those features to reference less flat is not factored in an explicit way. Ingenuity gains system then waits 3 seconds to ensure the ourselves.” We can look at the LRF data after the fact operational status as it helicopter has settled on the ground before Flight No. 5’s landing looked great, as and analyze it, but it’s not being used in relocates from Wright spinning down the rotors.” good as it could have been. Everything real time to navigate the copter.” Brothers Field. The downward-facing camera takes sev- went according to plan. “As we continue with our flights on eral images on landing, which is factored Then a momentous decision was made Mars,” Grip concluded, “we will keep dig- into the sequence for next takeoff. in Pasadena, to send Ingenuity further— ging deeper into the data to understand the into an operational demonstration phase, various subtleties that may exist and would SURPRISE ENDING very different, at a lower cadence for heli- be useful in the design of future aerial ex- AeroVironment’s role in fielding Ingenuity’s planned technological demon- copter operations. As the Mars Project fo- plorers. But what we can already say is: Ingenuity is built on a body of work that stration was to last for five flights. Then, cuses now on rover Perseverance and the Ingenuity has met or exceeded our flight dates back to its roots exploring human- sadly, its pathbreaking life would come to science it delivers, “We’re in a background performance expectations.”  powered flight, up through miniature products such as the aptly-named INSIDE Hummingbird and today’s HALE (high altitude long endurance) solar vehicles. For the Mars Helicopter Project, AeroVironment designed and developed the airframe and major subsystems, INGENUITY including its rotor system, rotor blades, and hub and control mechanism with AeroVironment hardware. The Simi Valley, California- based company also developed and built high-efficiency, lightweight propulsion motors, power electronics, landing gear, load-bearing structures and thermal by Abe Peck, Executive Editor enclosures for NASA/JPL’s avionics, Photo courtesy of NASA/JPL-Caltech. sensors and software systems. (The following “oral history” has been FIGURE 4 Feature tracks between a base image (left) and search image (right) during an outdoor test flight on Earth. edited and reordered for clarity.) CONTINUE TO “DESIGNING IT”

 www.insideunmannedsystems.com  June/July 2021 June/July 2021  www.insideunmannedsystems.com  49 COLLABORATION Mars Helicopter Project Mars Helicopter Project COLLABORATION

OPERATIONAL REQUIREMENTS uncontrolled. We didn’t realize that initially, tively benign compared to launching on a thermal vacuum testing, launch loads, put- PIPENBERG: The big challenge with a he- so after we learned that, then, “OK, where rocket, in terms of vehicle loads. The hub ting it on a shaker table and vibrating it, licopter like this is that you’re not setting a do we need to put a lot more time and effort reaction forces in flight are very small com- trying to simulate launch, as well as shock helicopter on the surface of Mars and then into the analysis? Where is it really going to pared to launching on the rocket. That’s re- testing, radiation and stuff like that. Those flying it. You’re really designing a small be paying off for us?” ally the challenge—you’re not just design- two helicopters were functionally similar standalone spacecraft that also happens That was kind of when JPL took on all of ing this to fly on Mars, unfortunately; we’re to the flight vehicle, and what we learned to fly. That really defines what a lot of the the avionics development, all the guidance, trying to design a very small, lightweight from those helicopters, we would roll into vehicle requirements are. navigation control, all the simulation mod- spacecraft that also happens to fly. the flight model. The environmental requirements were eling. For the most part, they handled all KEENNON: We also built something straight from JPL, from the Mars 2020 the batteries, charging solar array, all that. ITERATIONS that looked like a helicopter that was just mission: the Perseverance rover, mission, And AeroVironment took on the airframe, KEENNON: A dozen concepts on paper designed for testing the landing gear, the launch load requirements. With the launch a rotor blade design, propulsion motors, were never built. The first one [built] was shock absorbers, the angles. We did all this vehicle—an Atlas V—there’s a vibration servo swashplates, the primary structure. a box with a motor and rotors, but the first stuff in a motion capture room; it wasn’t a spectrum we have to survive that defines KEENNON: Then we went to the full- thing that looked like a helicopter. For the functional helicopter, but it was a function- most of the loads the helicopter is designed size demonstrator, the risk reduction, subscale, there were two versions that had al prototype for testing the landing gear. against. For controls, there are very spe- which was really a beautiful piece of engi- slightly different layouts, different passive PIPENBERG: At the end, it all came Early rotor test on the cific requirements: natural frequency, rotor neering. That was the first one that used stabilization systems. together. The airframe, the rotor system, Mars surface. blades for hubs, landing gear and all of that the full-sized rotor blades. These were all Then we went to the full-size aircraft. the landing gear, all of that was mated to flowed down from the JPL-GNC [guid- AeroVironment builds; the risk reduction There was one version we called the risk the avionics, the batteries, the solar array, ance, navigation, control] team. Sizing con- aircraft had the titanium parts that JPL reduction aircraft. And then we get into which was a big team effort. Solera de- straints was a huge one—what the vehicle is made. where Ben really was taking over the engi- veloped the solar cells, we developed the DESIGNING IT going to have to fit within. And that kind PIPENBERG: In addition to all those neering design models. structure they’re mounted on and it was BEGINNINGS of went both ways, right? AeroVironment [space and Mars] environments, it has to PIPENBERG: Two engineering devel- mated onto the rover through Lockheed MATT KEENNON [AV’s technical lead for the rotor system development on saying, “Hey, this is what we need.” And survive a test environment here on Earth. opment models were intended to be very Martin’s deployment system. A huge team Ingenuity and an AeroVironment principal electrical engineer] Lockheed Martin [its Mars Helicopter Before we ever get to Mars, we’ve already similar. One was used for flight testing in effort. AeroVironment started as a company making pedal-powered airplanes, Delivery System was designed to transport gone through almost the entire operational the JPL 25-foot-wide space simulator. The That final flight model had everything unusual man-carrying airplanes, a Solar Challenger. When I came on board in and deploy Ingenuity] and JPL working to life of the helicopter. Flying on Mars is rela- other was used for environmental testing: built to really survive on Mars. 1996, we were developing small surveillance aircraft for the military, including accommodate that, and also us working to a flapping-wing aircraft. We did a lot of DARPA work and DARPA (the Defense change the vehicle design to accommodate CONTINUE TO “READYING IT” Advanced Research Projects Agency) does the really far-out difficult things, not the space available. There was a lot of back knowing necessarily where it’s going to go. The Hummingbird was a basis for and forth on a lot of the environment sets. our weird and unusual flying machines that seemed to suit problem-solving for READYING IT unusual application, unusual requirements, really difficult customer requests. SURVIVABILITY JPL asked us, “How can we demonstrate flight in Mars easily?” and when I PIPENBERG: The first small-scale demon- JEREMY TYLER: [AV’s senior came on board, I brought in my ideas from those other projects. We did the sub- strators that attempted controlled flight in aeromechanical engineer; significant scale, and we also did this other one, which just had a motor and rotor blades. the chamber [December 2014] were rela- mechanical design work for most of the It went up and down on rails, it actually looked like the real helicopter, but it’s tively uncontrollable. Those initial small- metal and plastic components: servos, just basically an empty fuselage. But they all served their purpose of getting the scale helicopters were essentially built the linkages, swashplates, landing gear, excitement and interest levels raised to the next point of funding. same way we would build a helicopter that hinges, springs, latches] BEN PIPENBERG [AV’s engineering lead on the Mars Ingenuity Helicopter flies here on Earth. The problem we ended Thermal expansion was a huge chal- Program and AeroVironment senior aeromechanical engineer; worked on sub- up having is that, essentially, in a Martian lenge. The operating temperatures on Mars scale, larger demonstrators and the final build] atmosphere you have these rotors spinning are extremely cold, generally, but they’re Around 2012, 2013, the idea popped back up. The JPL chief engineer for the really fast, you have very low aerodynamic also very widely varying—it could be nega- program came back to AeroVironment and said, “Hey, what do you think about forces relative to the inertial forces in the ro- tive 100 Celsius, or it could be positive 20 Keeping it helping out with something like this?” That grew into small-scale risk tests, tor. And so the way the rotor system reacts Celsius. clean, with Ben essentially just putting a rotor system into a vacuum chamber representing a to control inputs is very different. You also But even that isn’t the full range. Every Pipenberg, Ingenuity Martian atmosphere, just demonstrating that we can generate lift. have much lower aerodynamic damping. engineering lead, component of the helicopter had to be Early on, there was doubt, even within NASA. There isn’t data out there Think of it like this really high-speed spin- and Sara Langberg, completely disinfected. For simple things, that tells you how this is going to work or what isn’t going to work. Coming up ning flywheel you’re trying to control with AeroVironment we could wipe them down or soak them in with tests that are relatively cheap and easy, and allowed to fail—that’s pretty very low aerodynamic forces, and there’s no aeromechanical alcohol. But partially assembled structures important. damping. So when you see these things fly- engineer. that might be damaged by alcohol, we had ing, they’re kind of all over the place, totally to bake them in an oven. So we had very

Photo courtesy of NASA/JPL-Caltech/ASU. Photo courtesy of AeroVironment.  www.insideunmannedsystems.com  June/July 2021 June/July 2021  www.insideunmannedsystems.com  COLLABORATION Mars Helicopter Project Mars Helicopter Project COLLABORATION

“But the big challenge with a helicopter like this is that you’re not setting a helicopter on the surface of Mars and then flying it. THE MEANING OF IT You’re really designing a small standalone spacecraft that also happens to fly.” On May 7, the Ingenuity Mars Helicopter

Ben Pipenberg, engineering lead and senior aeromechanical engineer, AeroVironment became an operational scout in addition to its original role as a technology dem- onstrator. Leaving both Wright Brothers Field and the Perseverance rover behind, high temperatures that the things had to had to account for that in almost all of the untwist and become flat. So it’s not only Ingenuity flew for 110 seconds, traveling survive, while also being able to withstand design aspects just to make it controllable. grabbing or counterweighting the whole 423 feet at a new height of 33 feet, cap- the low temperatures of service. It was a It was a very, very close collaboration be- blade, but it’s actually making the blade turing high resolution color images be- unbelievable range of requirements. tween the GNC folks and the mechanical torsionally stiff enough. That was an- fore landing at its new Red Planet home, Also, a large portion of the Perseverance folks. other driver that fed into the mechanical which bears the tepid but significant name mission is to look for signs of past life. This design of the blades, which again, comes Airfield B. is probably the most critical Mars mission so CHINESE WEIGHTS AND from our high altitude aircraft propeller PIPENBERG: It’s going to lead to more far in terms of biological cleanliness when TENNIS RACQUETS designs. capable science vehicles on Mars. In gen- A picture from the navigation camera it arrives. Every component had to be de- SARA LANGBERG [AV aeromechanical We don’t want to add mass to the blade, eral, there are a couple of options. There aboard Ingenuity captured the helicopter signed to be cleanable; every captured space engineer; testing, especially on but we can’t have them untwisting, un- are much larger helicopters that are being on takeoff during Flight No 2, showing on every captured component had to be ac- the blades; mechanical design and feathering. So, problems upon problems designed kind of as a rover replacement—a little sign of dust. cessible, such that it could be thoroughly fabrication, especially composites. Also would crop up, and then we would have big standalone science vehicle that can fly disinfected. Even from the mechanical de- swashplate geometry and building load to solve them in ingenious, lightweight, around, has instruments on board. There’s KEENNON: This thing is the first com- sign, we had to figure in disinfecting. models for the blades and how that clever ways. Our background doing these paired concepts with a lander like [2018’s] bination aircraft and spacecraft, and that KEENNON: And this is our first space interacts with the servos, plus landing other weird, lightweight, small aircraft, InSight—you have a smaller helicopter, is such an insane level of stress put on the project, so we had to add all this into our gear system in general] even like the Hummingbird, helped us or a quadrotor, or hexacopter, sort of like designers. plans and our budgets. They came out with When we’re spinning at such a high think about these things in a different way Ingenuity-size, that can go fly off and gath- PIPENBERG: It’s a good demonstration their gloves and petri dishes and swabs, RPM, different loads dominate than there and solve them. er samples and bring them back. And then of where we’re at in terms of autonomous inspecting, and taking samples from our would be on Earth. The blades are subject TYLER: On top of all that, the rotor you’ve got scouts for mapping missions, vehicles, in terms of spacecraft, rockets, work areas. It was intense. to three loads. There’s the aerodynamic blade is a very highly cambered airfoil, where a small helicopter paired to a rover deep space communications. NASA’s Ingenuity Mars Helicopter’s fourth load and an inertial load of the blade which is quite unusual for a rotorcraft, or or a manned mission to Mars, can gather We’re all extremely excited that flight path is superimposed here atop SYSTEMS CHALLENGES through its rotations. But there’s also the at least for a cyclic pitch rotorcraft. better imagery, or just operational aware- terrain imaged by the HiRISE camera Ingenuity’s success on Mars has allowed PIPENBERG: A lot of the features you can propeller moment, which is called “the ten- KEENNON: It’s inherently unstable. It ness of your surroundings. There’s a lot of aboard the agency’s Mars Reconnaissance NASA to transition the helicopter from a see on the helicopter were directly a result nis racquet effect.” It’s the tendency of the can’t fly for a half-second without making different ways to go. Orbiter. technology demonstration into an opera- of GNC requirements. For example, flying blades at an angle to try to flatten them- decisions based on the IMU and driving At AeroVironment, we are looking at tional demo phase. The increased range in a Martian atmosphere, there’s very low selves as they spin, due to having an off- the control system. We’ve tried flying this future technologies that can help: new WRIGHT BROTHERS ON MARS? and endurance demonstrated in the most damping; you’re essentially flying a gyro- axis center of gravity like we do. type of helicopter without a control system. materials, new fabrication methods, new LANGBERG: I think it’s very much a paral- recent flights can enable more advanced scope. And so, rather than having a huge On Earth, that’s a relatively small issue It goes bad quickly. It’s just the configura- designs, various types of mechanisms that lel Wright Brothers moment, in the sense missions, such as dedicated sensor plat- rotor system like what you would typically and doesn’t often need to be accounted for. tion of a coaxial helicopter. could be enabling for some of these mis- that we’re doing something that no one forms, mapping applications and sample have on a helicopter here on Earth, we But because we’re spinning five to 10 times TYLER: The coaxial router system sions. NASA has published a number of has ever done for the history of humanity. collection systems. made a very, very stiff rotor system. And faster on Mars, this effect really dominates. has no gyroscopic damping whatsoever. papers as well on kind of the Cadillac sci- Way back then, they were doing something We’ve had our fingers crossed that we with a counter-rotating, coaxial helicopter In order to get the servo loads down and Without a digital controller, without a ence mission; they’re really big standalone everybody thought was impossible. Look would be able to show that Ingenuity can like this, your net angular momentum is into a range where we can get good servo gyro-based autopilot, it’s entirely uncon- helicopters. So there’s a lot of interest, obvi- at where aviation has taken us now. So I provide a new dimension to planetary ex- therefore zero. response, we added these counterweights, trollable. ously, with a lot of excitement. And I think think there’s a lot of parallels into how this ploration on the surface of Mars, and this Making the rotor system really rigid sig- called “Chinese weights,” to the rotor. If you With all those aerodynamic loads of fly- the big hope here is that that generates re- is going to open similar doors, but for space next phase will allow us to do just that. nificantly simplifies how you can control look at the blades, there’s a hollow carbon ing and camber and twist, still the iner- ally good ideas. exploration. This is really a labor of love.  that helicopter. A lot of what you see on the fiber cone on each and a tungsten ball in tial loads were dominant. Sara’s analysis, helicopter in terms of even the rotor blade the tip that acts as a counterweight. That designing those counterweights to ideally shape, very high taper, relatively thick for counteracts the propeller moment effect, balance the inertial loads to the aerody- how low the Reynolds Number is on these and it doesn’t try to flatten itself quite so namic loads, let us really optimize the servo “…problems would crop up, rotor blades—that’s all for structural rea- much. They’re optimized to reduce the weight, and really minimize the size and and then we would have to solve them in ingenious, lightweight, clever ways.” sons. That really defines a lot of the reason servo loads the most at a hover condition. power that was needed for those servos to Matt Keennon, technical lead for rotor system development, AeroVironment the helicopter is built the way it is. So, even KEENNON: The blades are actually get the bandwidth that was required from though we didn’t work on the GNC side, we twisted, and they’re actually trying to the GMC folks at JPL.

CONTINUE TO “THE MEANING OF IT” Photos courtesy of NASA/JPL-Caltech/ University of Arizona and NASA/JPL-Caltech.  www.insideunmannedsystems.com  June/July 2021 June/July 2021  www.insideunmannedsystems.com  COLLABORATION Mars Helicopter Project Mars Helicopter Project COLLABORATION

Nawabi’s description of the he- into the rover and the spacecraft; one point of failure. And there are since a Kitty Hawk moment 100- licopter project sounded like a and sufficient autonomy to function so many other examples. The rover plus years ago. Achieving that as a “DARING ENDEAVORS” space-age Russian nesting doll. To far from Earth’s signaling. does very, very good things. But it team, across different locations dur- AeroVironment’s President/CEO defines the culture, teamwork get to Mars, it had to be specced to “And then,” he continued, “those does go slow. And its reach is very ing a pandemic—how could you not and execution behind the Mars Ingenuity breakthrough. fit within the Perseverance rover, requirements make it into a project limited.” be inspired? Just like we look now at which had to fit within the payload plan that says, ‘Here’s how I’m going Nawabi continued: “For every how aviation has transformed and “Their words, not mine, that this bay of the rocket that would carry to achieve that outcome.’ And then single potential mission in the fu- changed our lives, and humanity’s was the best, the best, collaborative them from Earth. Tested as an air- the work starts to be divvied out. ture, for not only Mars, for any other lives all over the world, for the last partnership that they could ever craft on our planet, the helicopter “You test it. And if it works, you de- planetary operations that mankind 100 years. The same could be said even wish for,” Nawabi recounted. “I functions as a spacecraft on Mars. velop it more, and you test it again, will take on—we believe it dramati- about this 100-plus years from now.” asked them, ‘Is there anything that And the fight control computer, the and you take the learning and you cally changes our ability to really One other aspect of collaboration we could have done better?’ ‘No, autopilot, had to be designed in keep improving it.” progress in exploration and eventu- needed to be addressed: how does Wahid, there’s nothing. We didn’t close coordination with JPL. ally habitating Mars with humans.” Nawabi see his role as a leader for even feel like a different team.’ The The stakes around Ingenuity MARTIAN MILESTONES Before concluding, Nawabi mir- such a complex project? collaboration was that close, that were higher than usual, Nawabi “On Mars, it’s a lot more difficult,” rored his earlier comment that “I am just here as an enabler to tight, that important. pointed out; after all, it’s harder Nawabi noted, citing another set of collaboration bookends all of remove the barriers, to create the milestones. “I’ve got a system that AeroVironment’s projects. environment and create the atmo- works in the lab. Would this sys- “The most important thing from sphere, so these people can unleash “Everything that we do tem become bulletproof under all the beginning,” Nawabi noted, “is and just basically live up to their starts and ends with collaboration.” these extreme conditions: radiation, that you have a team that is fully potential. We have a thing within shaking, vibration, rattle, extreme committed and believes in the out- AeroVironment that once you know “The credit goes to our team,” to fix something that’s 170 million temperatures? It will not fail, be- come of the mission. You’re going what you have to get done in terms Nawabi said. “I am just a messenger miles away. “You must be able to de- cause it cannot fail—the entire mis- to run into an enormous amount of of the mission objective or the cus- for them.” sign systems that 100% cannot fail,” sion’s over. uncertainty, most likely some set- tomer requirements, get the hell out Like a proud technological father, he said. “It’s not like you can launch “What if this thing lands, and it backs, challenges, questions, prob- of the way and let those people run. Wahid Nawabi, describing the collaborative success Nawabi defined the principles be- this thing and go return it and fix it. gets hit really hard; let’s test the lems. You can’t force that belief, it We let them work as free, collabora- of the Mars Ingenuity Helicopter Project. hind his company’s ability to deliver Every single intricate detail has to landing system, make sure it’s ro- has to come within themselves. tive, incredibly talented, incredible unique, even trail-blazing, solutions. be done just perfectly right. bust. The mechanical systems that “You take a daring challenge, and minds.” WAHID His tentpoles? Collaboration, a cul- “You’ve got to be able to work in have to open for this thing to be- a team of very talented people, both ture of innovation and mission-driv- an environment where there’s a come operational. Powering it up, at AeroVironment and at JPL—the —Abe Peck NAWABI en execution. tremendous amount of unknowns. the solar panels working—do we primary two parties, along with This has not been done before, so get enough energy to land and com- other players—essentially had a President/CEO, AeroVironment THE ROOT OF SUCCESS you have to figure out a way to do plete the mission? All these things moment that has not been repeated “Everything that we do starts and it yourself.” are very, very critical.” Wahid Nawabi, the head of multi- ends with collaboration,” Nawabi A multi-year project such as Six flights in, Nawabi turned the The team behind domain unmanned systems devel- explained emphatically. “When Ingenuity requires effective plan- proverbial telescope around to peer AeroVironment’s oper AeroVironment, was buoyant. you’re designing a system of sys- ning. Nawabi endorsed a PDCA into the future. work on the Mars 2020 Helicopter. Engineers and staff from NASA and tems solution, it is by far one of the model—plan, do, check, act—and “There already are several con- its Jet Propulsion Laboratory had most, if not the most, important, at- ticked off key milestones. cept missions that a flying device just visited to debrief in the wake tributes of the team and the entire “You develop a theory—what is this on Mars can do much better than a of the Mars Ingenuity Helicopter’s mission. concept, and what are we going to ground robot can do—that by itself shift from proof of concept to an “You can’t do a project like this achieve at the end of this? You check is a complete new dimension. For operational technology craft flying without having an incredibly strong the concept to make sure it’s valid.” example, samples on Mars in dif- above the surface of its namesake collaborative team” Nawabi contin- Next, Nawabi outlined, ‘you de- ferent places have to be meticulously thin-atmosphere world. As a key ued. “Taking things that have never velop a set of requirements as a picked up. You can use a rover, but partner, AeroVironment had devel- been done before is just what we result of that.” For Ingenuity, those what if the rover failed after taking oped components from the propul- thrive on. The people we hire, train requirements included the physics three samples? You can have an in- sion system to wiring and landing and develop have an appetite for of flight on Mars; size, weight and surance policy with a UAV. You don’t gear. daring endeavors. That’s No. 1. No. dimensions compatible with fitting risk a billion-dollar mission with 2, you’ve got to be willing to work passionately as a team. This is not a job. This is what they want to do.”

Photos courtesy of AeroVironment.  www.insideunmannedsystems.com  June/July 2021 June/July 2021  www.insideunmannedsystems.com 