Intent- and Fault-based Trajectory Prediction for Cooperative Localization and Collision Avoidance in Swarms

Abstract

Autonomous multi-agent systems similar to NASA’s CADRE mission could substantially improve the efficiency of robotic exploration on other planets. However, communication challenges pose a large risk to two main capabilities: collision avoidance and active collaborative localization. This work expands on Reachability-based Trajectory Design (RTD), intent prediction, and fault-based planning to develop an approach that improves the robustness of these capabilities under communication loss scenarios. This approach is then validated on turtlebots simulated in Gazebo.