Coordinated Unmanned Platforms for Autonomous Search Missions (CANINES)

Emergency response and source search applications demand efficient and accurate threat assessment. Traditionally, these efforts rely on manual surveys using handheld detectors or vehicle-mounted systems. Although effective, these methods can be enhanced by integrating radiation detectors and advanced contextual sensors onto autonomous platforms. In this work, a Spot robot is equipped with a radiation detector and a suite of contextual sensors, including LiDAR, an inertial measurement unit, and a visual camera, as well as an onboard computer capable of constructing 3D radiation maps of an environment in real time. Spot is first trained using reinforcement learning within the Unity game engine to develop navigation and exploration strategies. Once trained, Spot autonomously navigates outdoor and indoor environments, creating detailed 3D maps of these spaces, which are then overlaid with radioactive source localization data derived from radiation mapping and imaging algorithms. During field tests, Spot demonstrated successful autonomous navigation in unfamiliar environments, avoiding collisions while constructing 3D maps and accurately localizing radioactive sources. These results highlight the potential of autonomously navigating systems for mapping and source localization in diverse environments, paving the way for safer and more efficient emergency response operations.