四足机器人“盲爬”阶梯策略设计与实现

Design and Implementation of Quadruped Robots“Blind Climbing”Steps Strategy

针对四足机器人爬阶梯多采用基于零力矩点的静步态运动规划方法,存在攀爬速度慢、稳定性差的问题,设计了一种四足机器人快速“盲爬”阶梯策略。采用针对阶梯地形的轨迹规划摆动腿运动,基于广义动量法检测摆动腿与阶梯的磕碰事件,对磕碰事件进行了有效处理。根据本体感受器估计阶梯部分参数,结合稳定裕度概念规划机身的期望轨迹,采用模型预测控制对轨迹进行跟踪控制,保证了机器人快速稳定地攀爬阶梯。通过四足机器人实验平台验证所提方法的正确性,当机器人名义腿长为0.360 m、阶梯台阶高度为名义腿长的30.56%时,平均攀爬速度可达0.60 m/s(无量纲速度Fr达0.32)。

Aiming at the current situation that quadruped robots mostly use static gait motion planning methods based on zero moment points for climbing stairs,resulting in slow climbing speed and poor stability,a fast“blind climbing”steps strategy for quadruped robots is designed.The trajectory of the swing leg is planned for the step terrain,the collision event between the swing leg and the ladder is detected based on the generalized momentum method,and the collision event is processed effectively.The parameters of the steps are estimated according to robot’s proprioceptor.The torso’s reference trajectory is planned based on the concept of stability margin,then the trajectory is tracked and controlled by model predictive control(MPC),which ensures rapid and stable climbing.The correctness of the proposed method is verified by experiments on a quadruped robot.The average climbing speed can reach 0.60 m/s(equivalent to dimensionless speed Fr of more than 0.32)in the case the robot’s nominal leg length is 0.360 m and the step’s height is 30.56%of the robot’s nominal leg length.