APPL: Adaptive Planner Parameter Learning

While current autonomous navigation systems allow robots to successfully drive themselves from one point another in specific environments, they typically require extensive manual parameter re-tuning by human robotics experts order function new environments. Furthermore, even for just complex environment, a single set of fine-tuned parameters may not work well different regions that environment. These problems prohibit reliable mobile robot deployment non-expert users. As remedy, we propose Adaptive Planner Parameter Learning (appl), machine learning framework can leverage interaction via several modalities – including teleoperated demonstrations, corrective interventions, and evaluative feedback also unsupervised reinforcement learn policy dynamically adjust the classical response changes appl inherits safety explainability while enjoying benefits learning, i.e., ability adapt improve experience. We present suite individual methods unifying cycle-of-learning scheme combines all proposed performance through continual, iterative simulation training.