FLYING, HOPPING PIT-BOTS FOR CAVE AND LAVA TUBE EXPLORATION ON THE MOON AND MARS. J. Thangavelautham, M. S. Robinson, A. Taits, T. J. McKinney, S. Amidan, A. Polak, School of Earth and Space Exploration, Arizona State University, 781 E. Terrace Mall, ISTB4-795, jekan@asu.edu 1. Introduction: Wheeled ground robots are limited from exploring extreme environments such as caves, lava tubes and skylights. Small robots that utilize un- conventional mobility through hopping, flying and roll- ing can overcome many roughness limitations and thus extend exploration sites of interest on Moon and Mars. In this paper we introduce a network of 3 kg, 0.30 m diameter ball robots (pit-bots) that can fly, hop and roll using an onboard miniature propulsion system (Fig. 1). These pit-bots can be deployed from a lander or large rover. Each robot is equipped with a smartphone sized computer, stereo camera and laser rangefinder to per- form navigation and mapping. The ball robot can carry a payload of 1 kg or perform sample return. Our stud- ies show a range of 5 km and 0.7 hours flight time on the Moon. Fig. 1 – Pit-bot Cave Explorer Concept 2. Extreme Environment Exploration: High resolu- tion orbital imagery from LROC revealed evidence for subsurface voids and mare-pits on the lunar surface [1, 2]. Mare Ingenii shown in Fig. 2 is 70 m deep and is theorized to be collapsed entrance to a lava tube. The rugged terrain inside a lava tube entrance, with slopes steeper than 30o make it impassable by conventional Fig. 2 – Mare Ingenii wheeled robots. Accessible voids could be used for a future human base because they offer a natural radia- tion and micrometeorite shield and offer constant hab- itable temperatures of -20 to -30 oC [11]. Hopping bots [3-5] using mechanical systems are insufficient because of the expected rugged environ- ment particularly, when the slopes are too steep. Hop- ping poses challenges in determining where to land gently, particularly in rugged environment. In contrast flying allows for the systems to gently take off and land at a desired landing spot minimizing impact forces. Other methods such as tethering a probe to the base rover will not work in caves and lava tubes, because these formations are not straight, instead they are known to zig-zag. In addition tethers can catch on sharp rocks, displace rocks and risk tangling both the bot and the base rover. In contrast a flying robot is physically untethered to the rover and any risks it expe- riences leaves the rover unaffected. Current technology is severely limited by energy density of batteries and from miniature propulsion sys- tems [6]. These power constraints constrain mission duration, mobility and overall functionality of the small probes. To overcome the power problem we leverage advancements in miniaturized chemical mobility sys- tems together with integrating the required navigation and autonomous control technology into a small ball- shaped probe. 3. Flying and Hopping Pit-bots: Our proposed design consists of a network of 3 or more pit-bots (Fig. 3) to perform extreme environment exploration. The lower half of the sphere contains the propulsion system, with storage tanks for RP1 and hydrogen peroxide. Fig. 3 – Pit-bot Internals The attitude control system is in the top and contains micro-thrusters for maintaining yaw, pitch and roll. Next is the Lithium Thionyl Chloride batteries ar- ranged in a circle as shown. The mass budget is shown in Table 1. Comparison with other mobility options, including use of Radioisotope Thermal Generators (RTGs) and batteries show our design was found to be the only one to meet a minimal set of requirements (Table 2). A pair of stereo cameras and a laser range finder rolls on a turret enabling the pit-bot to take pan- oramic pictures and scan the environment without hav- ing to move using the propulsion system. Above the turret are two computer boards, IMU and IO-expansion boards and a power board. Table 1: Pit Bot Mass Budget Major Subsystem Mass (kg) Propulsion 1.2 Computer, Comms, Electronics 0.2 Power 0.3 Stereo Camera, Laser Ranger 0.3 Payload 1.2 Total 3 Table 2: Technology Comparison 4. Pit-Bot Propulsion: The critical subsystem required for this pit-bot is the propulsion system. The robots shall contain one primary lift engine positioned at the vehicles bottom portion and 8 “warm-gas” attitude control thrusters positioned at the top of the bot. For the ball robot propulsion we hereby consider RP1- H2O2 engine. Hydrogen-Peroxide is the oxidizer as well as the propellant for the Attitude Control System (ACS). Other oxidizers were considered for this robot including water, liquid-oxygen, and liquid nitrous- oxide. However, for the application of these small pit- bots, these oxidizers will not work. To begin, water may only be used to oxidize metal-hydrides and is not practical for use in an ACS since no source of heat is available to generate the required quantities of vapor. Liquid-oxygen requires cryogenic storage that is im- practical due to the size constraints of a 30 cm diame- ter vehicle within the lunar caves. Liquid nitrous-oxide requires immense pressures (7 MPa) for liquid storage and is quite difficult to ac- complish from a safety stand-point. Hydrogen peroxide is a good option because it can be tested at first with low purities (dissolved in water) to validate our physi- cal models and predictions. This minimizes risks dur- ing system development. Successful implementation at low purities will give us the confidence to increase to 50 % concentration. For a non-cryogenic fuel, RP-1 has by far the highest storage density of approximately 700kg/m3. Further- more, RP-1 is relatively low-cost, non-toxic, and easy to handle[7]. RP-1/H2O2 thrusters have been used since the 1960’s by the Soviet Union and have achieved TRL-9. However our efforts will be focused in minia- turizing the RP-1 H2O2 engine for the ball robot system (see Fig. 4). To implement this system in a small vol- ume and avoid the use of pumps and mechanical devic- es, our design uses pressurized nitrogen gas to initiate transport of the reactants into the combustion chamber. Prior to being injected into the main rocket-engine or the ACS valves, the hydrogen-peroxide is decomposed by means of a silver catalyst into oxygen and water. In the process of catalyzed decomposition, the oxygen and water will heat-up to a temperature of 600 oC. When the warm oxygen/water (oxidizer) is used to power the ACS system, the resulting specific impulse is approximately 180 seconds (no combustion). It is pre- dicted with this engine design, a specific impulse of 330 seconds will be achieved 50 % H2O2 concentra- tion. Fig. 4 – Pit-bot Propulsion System 5. Pit-bot Navigation and Mapping: The pit-bot would navigate by autonomously forming a triangular formation (Fig. 5, 6). The robots are equipped with bright lights that serve as beacon or as light sources in the lava tube/cave. Each robot moves forward, one robot at a time a short distance much like a bucket bri- gade [8-10]. Each robot takes stereo ground images, just before descending to the ground with one or both of the other robots in view. Because the ground robots have bright lights, a simple blob detection algorithm is sufficient to locate the ball robots in an image. Con- verting the stereo image to point cloud, provides dis- tance estimates to the ball robots on the ground. The robots will have sufficient computational capabilities to process stereo images. Using these distances, it is pos- sible to estimate the position of the ball robot relative to other robots on the ground (Fig. 7). Fig. 5- A network of 3 ball robots in a lava tube (1 flying mockup consists of a quad copter). Two on the ground are static display. Our studies show that the flying robot can locate/identify other robots within a 7 m distance. Fig. 6- The ball robots maintain a triangular formation, as each in order A, B and C take a short flight or hop to its next resting stop. Once they are in the designated triangle formation (inset 1), then lasers will be used to triangulate distance. If the other robots are visible, when all three robots are on the ground, the laser range finder could be used to get even more accurate distance measurements through direct triangulation. These positions would be recorded giving a total estimate of the position travelled by each robot from the base rover. Sections of a lava tube could be mapped (Fig. 8). Commercial point laser rangefind- er such as from Leica Disto E7100i have an error of 0.0025 % with a maximum range of 70 m. Using these estimates, the robots would be taking measurements every 9 to 5 meters interval. We would expect the total error in positioning using our approach to be 0.3 % to 0.5 % for 1 km radial distance. Fig. 7- Stereo images taken to produce 3D point cloud and mesh images of the pit-bots inside a lava tube (Flagstaff, Arizona). Fig. 8 - The pit-bots will obtain 3D images and pro- duce 3D maps of interiors as demonstrated in Gov- ernment cave (lava tube) near Flagstaff, Arizona. 6. Pit-bot Operations and Control: The pit-bots are intended to be fully autonomous. They will have the ability to hop, fly, hover, and roll. The robots will most often perform a fly-hop, which provides all the ad- vantages of hop, but with a soft landing. Optimal fuel saving trajectories have been found to obtain maximum hop range for given rocket engine specific impulse (Fig. 9). In addition, one of the goals of the pit-bot propulsion and attitude control system is to achieve hovering capability equivalent to current quad-copters. This hovering mode will be used to build 3D pa- naromic maps and for tracking the other pit-bots. A mission planner specifies the target coordinates where the ball robots and the payload package are delivered. At this point the pit-bots develop an internal navigation path avoiding obstacles in the path. In this approach each ball robot operates cooperatively, without a cen- tralized supervisor or leader to mitigate damage from loss of one or more robots. Fig. 9 – Comparison of pit-bot fly-hop trajectories to minimize fuel while maximizing range. 7. Conclusions and Future Work: Detailed concept studies of pit-bot design and use are ongoing. The re- sults to date show the principal feasibility of the navi- gation and controls approach. Development of an atti- tude control system also shows promising results. However, significant challenges remain in the devel- opment of the propulsion system even though the pro- pulsion technology is mature. The challenge will be in integration and miniaturization of the system into a 30- cm sphere. References: [1] M.S. Robinson et al., “Lunar reconnaissance orbiter camera (LROC) instrument overview,” Space Science Reviews, Vol. 150, No 1-4, pp.81-124, 2010. [2] R.V. Wagner and M.S. Robinson, “Distribution, formation mechanisms, and significance of lunar pits,” Icarus, Vol. 237, pp. 52-60, 2014. [3] S.B. Kesner, J.S. Plante, P.J. Boston, T. Fabian, S. Dubowsky, “Mobility and Power Feasibility of a Mi- crobot Team System for Extraterrestrial Cave Explora- tion”, IEEE International Conference on Robotics and Automation, 2007 [4] P. Boston and S. Dubowsky, “Hopping Mobility Concept for Rough Terrain Search and Resue,” AIAA Space Conference, 2006. [5] M. Pavone, J. Castillo-Rogez, J. Hoffman and J. Nesnas, “Spacecraft/Rover Hybrids for the Exploration of Small Solar System Bodies,” NASA NIAC Phase I Final Report, 2012. [6] J. Thangavelautham and S. Dubowsky, “On the Catalytic Degradation in Fuel Cell Power Supplies for Long-Life Mobile Field Sensors.” Journal of Fuel Cells: Fundamental to Systems, pp. 181-195, 2013. [7] R. Arnold, P. Santos, T. Kubal, O. Campanella, W. Anderson "Investigation of Gelled JP-8 and RP-1 Fuels," WCECS, 2009 [8] J. Thangavelautham, K. Law, T. Fu, N. Abu El Samid, A. Smith and G.M.T. D’Eleuterio, “Autono- mous Multirobot Excavation for Lunar Applications,” Robotica, pp. 1-39, 2017. [9] J. Thangavelautham, N. Abu El Samid, P. Grouchy, E. Earon, T. Fu, N. Nagrani and G.M.T. D'Eleuterio, “Evolving Multirobot Excavation Controllers and Choice of Platforms Using Artificial Neural Tissue Controllers,” Proceedings of the IEEE Symposium on Computational Intelligence for Robot- ics and Automation, 2009, DOI: 10.1109/CIRA.2009.542319. [10] J. Thangavelautham, A. Smith, D. Boucher, J. Richard and G.M.T. D'Eleuterio, "Evolving a Scalable Multirobot Controller Using an Artificial Neural Tissue Paradigm," Proceedings of IEEE International Confer- ence on Robotics and Automation, 2007. [11] G. Heiken, D. Vaniman and B. French, “Lunar Sourcebook: A User's Guide to the Moon,” Cambridge Univ. Press, 1991 .