Cooperation and Coordination Principles for Icy Moon Exploration with Autonomous Multi Robot Systems

This paper presents research motivated by a higher-than-ever level of multidimensional autonomy of planetary su-rface exploration beyond the Mars orbit required from robotic systems. Obviously, this is due to the long time duration requi-red to establish communication with the nearest human opera-tor and the expected high complexity of astrobiological explo-ration of moons hosting a subsurface ocean (“Icy Moons”). First, we present a summary of challenges and bottlenecks related to the autonomous exploration of Icy Moons. Next, we characterize the software requirements necessary for the reliable deployment of teams of autonomous planetary surface robots. This ensures higher exploration robustness compared to the standard systems using single units. Third, we address the need for new concepts for multi-level robot decision and learning autonomy as well as novel approaches to anticipatory coordination and cooperative phenomena in robot teams. Specifically, we propose an application of anticipatory robotics principles in planning the cooperation of deep space intelligent exploratory robots with autonomous decision making capabi-lities and knowledge sharing of other robots' decision algorithms. This latter knowledge sharing is essential in situations where communication is limited during the execution of a sha-red task to predict future actions of the robotics system. Cooperating robot formations are modeled as evolving antici-patory networks, driven by discrete event systems with virtual supervisors. The computational homogeneity among the robots allows each of them to periodically act as a coordinator and issuing commands to other robots while integrating their anti-cipated positions and activities. The above principles have been validated in experiments with inspection and harvesting robot teams. We will also propose new solutions to travel path plan-ning for multiple exploratory rovers based on an additive transformation of probabilistic risk superposition on path segments. As an illustrative example, we present a multicriteria coordination problem involving a team of robots planning non-dominated paths and collaborating in an anticipatory network. Finally, we assess planned deep space missions to identify pro-mising applications for coordinated anticipatory robot teams.