融合几何约束特征提取与多奖励协同优化的仿生步态控制

连远锋,范树玉,张本哲,王森

Bio-inspired gait control integrating geometric constraint feature extraction and multi-reward cooperative optimization

Yuanfeng LIAN,Shuyu FAN,Benzhe ZHANG,Sen WANG

表 1 四足机器人仿真训练中不同算法的计算效率与碰撞性能对比

Tab.1 Comparison of computational efficiency and collision performance of different algorithms in quadruped robot simulation training

算法名称 $ {t_{\text{e}}} $/s $ {c_{\text{e}}} $ walk-these-ways-go2 6.11 7.51 SAC 6.05 9.19 所提方法 6.83 3.76