<<

ICES-2020-51

Mapping CO2 Within A Spaceflight Analog Environment Tristan C. Endsley1, Theodore J. Steiner, III2, Forrest E. Meyen3, Kevin R. Duda4 The Charles Stark Draper , Cambridge, MA, 02139 USA

Marcum L. Reagan5 NASA Johnson Space Center, Houston, TX, 77058, USA

Carbon dioxide (CO2) levels onboard the International (ISS) have been reported to be as high as 10 times greater than ambient levels in a terrestrial environment, which can harm crew performance and productivity. NASA’s personal CO2 monitor provides ISS with the ability to monitor the CO2 levels, , and via a body-worn system; however, it lacks localization to determine where the sensor measurements were taken. A CO2 sensor was integrated with the Draper Wearable Kinematic System (WKS)— a self-contained, wearable, and hand-portable system for real-time navigation state estimation. The system was demonstrated operationally within the NASA Extreme Environments Mission Operations (NEEMO) mission 23. The NEEMO missions are conducted in the Aquarius Base underwater research facility, located 19 m below sea level off the coast of Florida. This Earth-based spaceflight analog environment creates a realistic platform through which to examine human performance and simulate operations that are representative of living and working in space. The WKS – assembled primarily from commercial-off-the-shelf equipment – analyzes the monocular vision and inertial measurement unit data to generate a real-time navigation state estimation utilizing the Draper smoothing and mapping with inertial state estimation (SAMWISE) algorithm. The WKS+CO2 sensor system tracked crew position, velocity, and orientation while mapping CO2 concentrations within the underwater as a function of time during normal daily crew activities. The results of that testing are discussed, and challenges associated with data collection in this are summarized. This capability can provide astronauts, flight directors, and ground support personnel with a better understanding of environmental conditions to improve task efficiency, crew productivity, and appropriate cycle times and operations of the environmental control and life support systems (ECLSS) onboard the ISS.

Nomenclature

CO2. = Carbon Dioxide COTS = Components Off The Shelf ISS = International Space Station WKS = Wearable Kinematics System MCA = Major Constituent Analyzer ECLSS = Environmental control and life support systems

1 Senior Human Systems Engineer, Human Systems and Visualization Group, Systems Engineering Directorate, 555 Square Cambridge, MA 02139. 2 Principal Member of the Technical Staff, Perception and Localization Group. 555 Technology Square Cambridge, MA 02139. 3 Senior Space Systems Engineer, Space & Mission Critical Systems, 555 Technology Square Cambridge, MA 02139. 4 Principle Space Systems Engineer, Group Lead, Space & Mission Critical Systems, 555 Technology Square Cambridge, MA 02139. 5 NEEMO Mission Commander, Mission Planning Office, NASA Johnson Space Center, Houston, TX, 77058.

Copyright © 2020 The Charles Stark Draper Laboratory, Inc. EKF = Extended Kalman Filter MSCKF = Multi-State Constrained Kalman Filter SAMWISE= Smoothing and mapping with inertial state estimation NHV = Net Habitable Volume NEEMO = NASA’s Extreme Environment Mission Operations USOS = United States On-orbit Segment EVA = IVA = Intravehicular Activity IMU = Inertial Measurement Unit GPS = Global Positioning System FLA = DARPA’s Fast lightweight Autonomy 6-DOF = 6 degree-of-freedom

I. Introduction

levated CO2 levels within the ISS environment are a potential source of acute headaches, dizziness and high blood 1 E , among other serious symptoms . While CO2 expiration is a natural part of the human process (humans generate CO2 at a rate of 0.9-1.2 kg per day), build up and retainment of CO2 gases within the ISS 2 environment can lead to crew performance and productivity degradations . Additionally, CO2 accumulates non- uniformly within the habitable ISS environment, with CO2 concentrations averaging 0.5±0.2% (2.3-5.3 mm Hg) versus terrestrial levels of 0.03% by volume (0.23 mm Hg)3,2. Large variations in the experienced 7-day averages have ranged from 3.39 mm Hg to a peak of 4.50 mm Hg 3,1. Described by James et al. (2010), “…at times and in certain locations CO2 levels can go well above the levels measured by the MCA [Major Constituent Analyzer]. If crewmembers are working in a location with suboptimal airflow, then local concentrations at the breathing zone can be somewhat higher than the module average”2.

4 The NASA Personal CO2 Monitor is capable of quantifying CO2 levels up to 5% by volume. It has already successfully demonstrated the identification of CO2 fluctuations in the ISS habitat environment. The NASA WEAR lab used the device to identify a non-homogeneous dispersion of CO2 gases, which were found to develop high pockets in the ISS. There is, however, no location data associated with these measurements. Despite being able to record nearly continuous measurements of the carbon dioxide in the ISS atmosphere, there is no data to determine where the measurement was taken.

Other systems like the Major Constituent Analyzer (MCA)—a mass spectrometer integrated into the ISS— which is primarily used in the United States On-orbit Segment (USOS), only collects data at multiple fixed points in the 5 habitable cabin volume . While this allows highly accurate and continuous measurements of CO2 in the ISS to be collected, the “MCA…can only measure gas constituency at fixed points which are plumbed for sample taking”5 This leaves a void in the ability to map the complete CO2 environment of the ISS to inform operation and cycle times of the onboard Environmental control and life support systems (ECLSS) appropriately. The habitable volume of the ISS 3 5 (790 m ) does not generate uniform dispersion of CO2, leading to high concentrations within some parts of the environment. Other types of platforms, such as those that are affixed to the ISS surface structure are equipped to detect the boundary conditions of the habitation environment, however the lack of convection in space will mean that interior environments are not adequately monitored. Measurement and localization of CO2 concentration is necessary to understand how gases collect and disperse in microgravity and is essential to ensure safety in long-duration space .

Our research demonstrates the ability to provide a time-based location associated with each CO2 measurement via a CO2 sensor integrated with Draper’s vision+inertial navigation system—the Wearable Kinematics System (WKS)6,7. The WKS is a generic Figure 1: The Draper Wearable location that can be used with many sensors and many Kinematics System 2 International Conference on Environmental Systems

investigations. In this paper, we demonstrate the use of the Draper wearable kinematic system (WKS) for the localized detection of CO2 in the NEEMO Habitat environment, as a proof of concept for the utilization of such technology for the detection of gases and other environmental elements within the ISS habitat (and other environments).

II. Background The WKS collects critical data on crew position, orientation, and navigation states within a microgravity environment6,7. As a personalized position and state estimation unit, the WKS can capture critical data on crew activities over time and can quantify crew use of space for the context of informing habitat usage without the use of GPS or affixed structures. The Draper vision + inertial navigation device is a generic technology that provides location and orientation information to any sensor or system integrated with it. This study augments the platform through the integration of a CO2 sensor.

A. The Draper Wearable Kinematics System The Draper Wearable Kinematic System (WKS) is a wearable device that estimates the wearer’s position and attitude in dynamic, GPS-denied environments. Draper algorithms calculate position estimates by incorporating measurements from an Inertial Measurement Unit (IMU) and a camera that both reside inside the WKS unit. Vision+inertial navigation systems are a versatile and flexible approach for metric navigation and map building for a variety of GPS-denied applications, such as small air vehicles operating indoors, or personal body-worn navigation systems. These systems, at their core, include a 6 degree-of-freedom (6-DOF) IMU, imaging system, and may also include a baro- altimeter pressure/height sensor. (For operations in Figure 2: Demonstration of WKS Location the ISS microgravity environment, a baro-altimeter tracking on the ISS mock up will not provide useful information). Our previous work6 utilized a Multi-State Constrained Kalman Filter (MSCKF), a vision-aided inertial navigation algorithm proposed by A. Mourikis, et al8-10. In the MSCKF, the vision and IMU measurements are integrated into a single Extended Kalman Filter (EKF) as opposed to a hierarchical or cascaded filter approach. Several key advantages of the algorithm are that it handles nonlinear visual measurements better than the standard EKF, it is a consistent linearized estimator where navigation uncertainty is accurately represented, the sliding window filter approach allows multi-rate sensor data, and it estimates the camera intrinsic and camera-to-IMU extrinsic calibrations online and therefore eliminating the need for high-precision pre-calibration routines.

Draper’s Smoothing And Mapping With Inertial State Estimation (SAMWISE) algorithm provides a flexible framework for real-time sensor fusion and aided inertial navigation6. SAMWISE fuses data from a primary inertial measurement unit (IMU) with data from any available combination of aiding sensors to produce a robust, high-rate, high-accuracy position, velocity, and orientation (state) estimate. This estimate is suitable for real-time vehicle control or personnel tracking. SAMWISE utilizes a sliding window incremental nonlinear smoother, as opposed to a more traditional Extended Kalman Filter-based (EKF) approach, which achieves more efficient information extraction from highly nonlinear sensor measurements, such as visual feature or landmark tracking. SAMWISE can perform multiple Gauss-Newton iterations and re-linearize the nonlinear visual feature measurements according to the optimal state values, as opposed to the state prior to adding the measurements. The result is a vision-aided inertial navigation system that achieves high-accuracy while maintaining real-time operation on embedded hardware12. SAMWISE has performed sensor fusion using various combinations of sensors, to date including an IMU, EO/IR camera, GPS, laser and barometric altimeter, magnetometer, and scanning lidar, and incorporates internal health monitoring capabilities to detect abnormalities or failure conditions.

3 International Conference on Environmental Systems

SAMWISE uses a Bayesian factor graph to represent state estimation problems, which is particularly well-suited for probabilistic optimization problems and trivially extendable to any combination of supported sensors. Estimating the vehicle trajectory with a sliding window in real-time incurs a slight processing delay, typically averaging around 200ms. To overcome this delay and consistently produce state estimates at the high rates necessary for high-rate control of dynamic flight systems, SAMWISE incorporates a low-latency strapdown inertial propagator to output position, attitude, and velocity estimates at 100+ Hz with less than 1 ms latency. SAMWISE additionally uses a novel measurement buffering approach to seamlessly handle delayed measurements, measurements produced at inconsistent rates, and sensor data requiring significant processing time, such as camera imagery11. This methodology allows factors to be added to the system at a delay, enabling techniques like delayed initialization of nonlinear variables and outlier checking, or for outlier measurements to be removed after they are added to the system.

B. WKS and CO2 Hardware

The WKS is a modular hardware system that integrates a computer, sensors, power, and human interfaces into a single streamline package. The system is worn across the chest and held close to the body with two shoulder straps and a waist strap. The geometry of the housing minimizes system volume and is angled to minimize intrusion into the user’s usable work volume. The structure hosts an Intel NUC mini PC for executing algorithms and recording data. Cooling vents on the slanted face of the housing direct warm air away from the operator. The Draper IMU sensor board is mounted firmly at the bottom of the housing and is mechanically coupled to the camera mount location. The Point Grey Flea3 protrudes from the front of the device to maximize field of view. A power switch on the front of the device initializes the primary computer. An LCD display mounted on the top of the housing is easily viewable by the operator and shows system status. A data logging switch to the right of the display is used to initialize data collection. An internal battery is attached to a removable back panel. The battery is charged through an external charge port on the left side of the device. Data can be off-loaded through Wi-Fi or the external Ethernet port on the right side of the housing. The CO2 sensor is mounted to the right panel of the device. The CO2 sensor is the CozIR by co2meter.com. The sensor is a Non-Dispersive InfraRed (NDIR) sensor. It is small sensor weighing only 20 grams with an average power consumption of 1.5 mA @ 3.3V. The sensor has a resolution of 1 ppm and an accuracy of +/- ppm +/- 3% of reading and a pressure dependence of .13% per mm of Hg. The right and left panels are removable to enable different sensors to be easily swapped and tested.

Figure 3: WKS Hardware Components

C. Prior Work Prior to its deployment at NEEMO, the underlying vision+ inertial navigation hardware and SAMWISE algorithm had been evaluated for multiple applications such as Person tracking, including astronauts on ISS (NASA WKS 4 International Conference on Environmental Systems

Program), high-speed quadrotor flight in GPS-denied environments (DARPA FLA Program), Augmented reality for training or industrial applications (Draper Monarch Program), and Autonomous ground vehicle tracking in multiple test environments, including the ISS mock up facility, the NASA HERA facility, in warehouses, office buildings, college campuses, forests, runways/airports, open fields, rural and urban streets, and ski trails.

D. NASA Extreme Environments Mission Operations (NEEMO) Analog In 2019, Draper participated in the NASA Extreme Environment Mission Operations (NEEMO) 23 spaceflight analog mission. The analog environment of NEEMO facilitates an operational environment more closely aligned to the conditions inherent to space flight operations. NASA’s underwater exploration analog brings together astronauts, mission operations, and marine scientists to explore and exploration objectives related to spaceflight missions to the ISS, Moon and . crew simulate living on , living in isolation, and testing exploration spacewalking techniques. NEEMO provides the opportunity to explore scientific or operational objectives, test spaceflight with representative populations and conduct marine science during the course of engineering saturation test week and the 9-10 day mission. In operation since 2001, NEEMO takes place in the Aquarius Habitat, which is located on the floor 62 feet (19m) below the surface and 5.4 miles off the coast of at . The Aquarius habitat is an undersea laboratory originally built by NOAA, now owned and operated by Florida International University. A crew living inside Aquarius is “saturated” at that depth, which allows up to 9 hours/day of dives from the habitat. Boat based divers would be challenged to get 1-2 hours of meaningful work in the same depth range. The ARB is a high-humidity and high-pressure environment known to cause challenges for COTS hardware. Humidity is kept around 75%, and fluctuates +/- up to about 3 percent. Pressure is ~ 48 fsw, which is ~ 2.4 ATM. Tidal changes and waves cause that to fluctuate by a few feet. NEEMO Mission 23 simulated a Lunar Exploration Analog.

NEEMO creates a realistic mission environment through which to examine human performance, with representative astronaut populations, across dynamically changing conditions, and allows for the examination of highly critical operations in situ. During the course of the engineering saturation test event, we collected several data sets using the WKS and CO2 sensor, which provided in situ data points for changes in the environmental CO2 state within the ARB habitat based on crew behaviors (i.e. discussions in the bunk area) as localized to the position of where the sensor data was collected.

As a secondary objective, the WKS was utilized to localize and provide position estimation of the four-aquanaut crew at different times during the 9-day mission, during the course of a variety of activities. This objective aims to quantify the net habitable volume (NHV) requirements of spaceflight habitats by providing quantitative data on the crew utilization of space during mission operations. During the course of the NEEMO mission, the crew performed free roam tasks, as well as different types of mission tasks such as collecting supplies to perform another experiment across the habitat, and conducting IVA activities while wearing the WKS device.

Figure 4: The Aquarius Reef Base (ARB) Habitat Layout

III. Current Study

Capturing measurements of CO2 states within habitable spaceflight environments longitudinally over a mission period will provide critical insight into the environmental cycling requirements for the maintenance of effective and safe crew operations. Additionally, tying the localized CO2 data to specific human behaviors will be a fundamental 5 International Conference on Environmental Systems

next step in the identification of appropriate crew task allocations and behaviors within the ISS habitable environment, and can help inform decisions about atmosphere management, dynamic task allocation, and crew performance.

In the current study, the WKS was deployed with a Commercial off the shelf (COTS) CO2 sensor— the C2IR— to demonstrate the capture and localization of CO2 distributions over the duration of a crew task within the NEEMO mission 23 analog test environment. Our CO2 sensor was factory calibrated to be accurate to 50 ppm within the range of 0 to 10,000 ppm (CozIR GC-0012) and Aquarius is within specified operating conditions. We would not expect a meaningful sample below 50 ppm. However, it is important to note that our CO2 sensor was not calibrated within the operational or pressure environment of the habitat, and as such the results presented are not representational of the absolute values of CO2 during the mission, rather they are illustrative of the localization of exemplar measurements.

Because Aquarius is located in an extreme environment, its life support systems are critical to maintaining a habitable environment. Consequently, there are a number of onboard life support parameters (e.g., pressure, PPO2, PPCO2, etc.) that are continuously monitored onboard and at the shore base where a watch is kept 24/7 during a mission. Air is circulated through the cabin where it is pulled into CO2 absorbent beds and returned to the cabin. In addition, during nominal ops a small rate of fresh air is continuously supplied, meaning that a small rate is continually being vented overboard through the Wet porch (far right of figure 4). The CO2 absorbent is a consumable, and once it starts to become saturated CO2 rises in the cabin, indicating that the beds need to be changed. This happens every ~ 4-5 days during a mission. Once the absorbent beds are replaced, CO2 quickly falls back to the normal control level.

A. Data Collection The tests occurred during the Engineering Saturation test dive, where mission management conducts tests of all equipment and technology prior to the mission. Following an initialization phase, the WKS operator donned the system via a chest mounted strap. They were instructed to wear the WKS system while carrying out their normal mission tasks. Data from a 10 minute and 31 second data collection period is presented in Figure 5 below. Over the course of the data collection session, the operator conducted multiple walks throughout the habitable space. This involved multiple revolutions throughout the habitat, and included forward movement, translations, stops, and periods discussions with other members of the team.

IV. Results

Over the time course of the data collection period, fluctuations in the concentrations of the measured CO2 within the habitat were captured (Figure 5). Trends in concentration levels were observed to increase over the general duration of the session, as approximately 4 people congregated in certain areas, or as the operator moved to different areas where air flow in the habitat was greater.

1. Mapping Operator behaviors and localized CO2 measurements in the habitat. Localization of CO2 measurements are demonstrated in figure 6 below. Initialization and Concentration completion of the collection occurred at the table 2 (near the center of Figure 6). The representational CO values of CO2 are visualized via a heat map, where Green: Instantaneous reading in lower levels of measured CO2 are in blue and Blue: Filtered reading relatively higher levels are indicated by darker red.

Time (s) Over the course of the data collection session, the operator conducted multiple walks throughout the Figure 5: Observed CO Measurements over time 2 habitable space. Near the end of the trajectory, three people were chatting in the bunk room in close proximity, which may have caused the rise in local CO2 levels, through the combined metabolic activities of the crew in the enclosed area.

6 International Conference on Environmental Systems

Figure 6: Heat map of CO2 concentrations in the ARB habitat

As the WKS operator moved towards the Wet porch (far right of Figure 6), lower levels of CO2 were measured closer to air flow outputs. Accuracy of localized measurements are within 50 cm.

V. Conclusion The integration of the Draper Wearable Kinematics System (WKS) within NEEMO provides a valuable demonstration of the WKS as a system for tracking astronaut position, velocity, and orientation within an operational, GPS- denied analog environment. Knowing accurately where a crewmember is and what is going on in the environment around them has numerous applications to NASA—whether that crew is conducting operations inside a spacecraft or habitat (such as the ISS, Gateway, one day the Mars Transit Vehicle), or an EVA on a planetary body. Providing accurate information about crew position, orientation and navigational states and environmental conditions within and around spaceflight habitats (ISS, HLS, Lunar Gateway)– via various sensors and modalities – will provide critical insight to astronauts, flight directors, and ground support personnel.

The WKS estimates the crewmember’s navigation state vector – position and orientation – as a function of time during the course of their normal daily activities. The ability to convey this data as a function of time and location has implications for the characterization and quantification of position and navigation estimates from the astronauts living and working within exploration habitats, like the ISS and Lunar Gateway, and in analog habitats like Aquarius Habitat; and provides a unique opportunity to inform the design of future spaceflight vehicles, and supporting subsystems, as well as provide accurate and active tracking of crew position and orientation.

In this study, measurements of environmental CO2 were captured and mapped to the habitat based on crew’s navigational state vector. Systems designed for reducing metabolic byproducts of astronaut crew such as CO2 can be improved by having an accurate map of the source location and quantity of these byproducts. NEEMO provides significant value as a spaceflight analog, simulating extreme environmental conditions similar to a spacecraft or ISS application. As in spaceflight environments, NEEMO creates a very real extreme environment, in which every minute humans are alive is due to mechanical life support systems. The ability to easily monitor how life support systems are performing, and to intelligently cycle those systems requires an ability to effectively localize sensor and crew positional data. ECLSS system parameters (flow rates, location, etc.) can be tuned to environmental usage patterns. A real-time system may even be able to take this information and adapt in real time by directing habitat circulation and enhancing flow rates to particular areas of the habitat. This data can be used to provide a better understanding of the environmental conditions to improve task efficiency, crew productivity, and appropriate cycle times and operations of the environmental control and life support systems (ECLSS) onboard the ISS, and other spaceflight habitat environments.

The WKS is a reliable, accurate way for real time navigation state estimation of crew members that not only track and report the current position of the astronaut, but also provides an integrated sensor capability that tie operation position to environmental states. NEEMO Mission 23 validated the WKS within a spaceflight analog, demonstrating purpose to numerous applications within in-space vehicles. Future work to prove and demonstrate the applicability of this technology for personal navigation and position estimation on a planetary surface is in progress. 7 International Conference on Environmental Systems

Acknowledgments This research was supported through Internal Research and Development funding from Draper. The Draper team would like to thank the NEEMO mission management team, the NEEMO aquanaut crew the FIU Aquarius Reef Base crew for their support in the collection of this data.

References 1Law, M. Van Baalen, M. Foy, et al., "Relationship between carbon dioxide levels and reported headaches on the international space station," Journal of occupational and environmental medicine, vol. 56, 2014, pp. 477-483. 2 James, J., Matty, C., Meyers, V., Sipes, W., & Scully, R. “Crew health and performance improvements with reduced carbon dioxide levels and the resource impact to accomplish those reductions.” In 41st International Conference on Environmental Systems, p. 5047, January 2011. 3 A. Stankovic, D. Alexander, C. Oman, et al., "A Review of Cognitive and Behavioral Effects of Increased Carbon Dioxide Exposure in Humans." NASA Technical Paper. 2016 4 NASA. Personal CO2 Monitor. Available: https://www.nasa.gov/mission_pages/station/research/experiments/2101.html (2017, November 29). 5 Matty, Christopher. "Overview of carbon dioxide control issues during international space station/space shuttle joint docked operations." In 40th International Conference on Environmental Systems, p. 6251. 2010 6 K. R. Duda, T. Steiner, and P. A. DeBitetto, "Ground-Based Performance Evaluation of a Wearable Vision+Inertial Navigation System for ISS Net Habitable Volume Estimation " presented at the IEEE Aerospace Conference, Big Sky, MT. 2017 7K. R. Duda, R. A. DeFronzo, T. J. Steiner, Chamitoff, G., "Development of a Wearable Vision+Inertial Navigation System for International Space Station Intravehicular Activity Operations " in 47th International Conference on Environmental Systems, Charleston, SC. 2017 8 M. Li and A. I. Mourikis, "High-precision, consistent EKF-based visual–inertial odometry," The International Journal of Robotics Research, vol. 32, pp. 690-711. 2013 9 A. I. Mourikis and S. I. Roumeliotis, "A multi-state constraint Kalman filter for vision-aided inertial navigation," in 2007 IEEE International Conference on Robotics and Automation, pp. 3565-3572. 2007 10 A. I. Mourikis, N. Trawny, S. I. Roumeliotis, et al., "Vision-aided inertial navigation for spacecraft entry, descent, and landing," IEEE Transactions on Robotics, vol. 25, pp. 264- 280. 2009 11 T. J. Steiner, R. D. Truax, and K. Frey, "A vision-aided inertial navigation system for agile high-speed flight in unmapped environments," in Proce. IEEE Aerospace Conference. 2017 12T. J. Steiner, T. C. Endsley, K. R. Duda. “A Loop Closure Hierarchy to Improve the Robustness of a Wearable Vision+Inertial Navigation System,” IEEE Aerospace Conference Yellowstone Conference Center, Big Sky, Montana, Mar 3 - Mar 10, 2018

8 International Conference on Environmental Systems