Data Sparse Inference of Operator Spatial Reward Models in Uncertain Environments

Hunter Melchior Ray^*Aditya PandeyNisar Ahmed

• University of Colorado Boulder, Boulder, United States

Human-machine teaming allows people to leverage the impressive capabilities of autonomous robotic teammates to safely accomplish challenging tasks. While users may be experts in their fields, robotic interfaces need to be intuitive to the general population and able to quickly interpret minimal user input from multiple modalities in directing autonomous teammates towards key locations for information-based tasking. This work presents a flexible multimodal algorithmic and visual interface that enables dynamic reprogramming of autonomous planning algorithms, which we focus on the use of Uncrewed Aerial Systems engaged in outdoor Search and Rescue. The Responsive Interface for iNtuitive Aircraft Operation (RINAO) leverages known geographic database information, such as trail networks, in conjunction with a variable set of user-defined features, such as search areas and landmarks, to efficiently infer a mission-specific, uncertainty aware geospatial interest distribution that informs optimal planning algorithms through reward shaping. The approach is validated using ten experts in public safety with 13.5 years of median operational experience. Results of this user evaluation show that the system enables effective and efficient alignment of geospatial interest and above average usability. Evaluating the system's performance against an Inverse Reinforcement Learning baseline, we find that our approach meets or exceeds the baseline's value alignment while performing inference with substantially less time and user input.