arxivcs.ROcs.LG2026-07-17
Learning Reach-Avoid Task with Reinforcement Learning: Vectorized Simulation and Benchmark
Deep reinforcement learning (DRL) has a longstanding tradition in addressing the reach-avoid task problem, especially for controlling robotic arms. While this task serves as a baseline environment within the research community, the ability of DRL to effectively learn the each-avo…