PPO 会探索、CBF 会保命:RADAR-BPO 把复杂环境碰撞率压到 10.7%原文:Risk-Aware Reinforcement Learning with Dynamic Safety Filter for Collision Risk Mitigation in Mobile Robot Navigation 作者:Bingbing Guo、Guina Wang、Yiyang Chen、Yue Gao(苏州大学机电工程学院)、Qian Xie(江苏易程科技股份有限公司) 期刊:Sensors 2025, 25, 5488(MDPI,2025-09-03 出版,DOI: 10