重载四足机器人的足部与地面接触过程和步态转换过程中会受到不确定的冲击载荷作用，易导致足部机构载荷过大从而造成结构的冲击损坏。因此，针对使用液压串联弹性执行器（series elastic actuators，SEA）作为足部末端在非结构环境下动态性能差的问题，提出了基于环境参数估计的滑模阻抗控制方法（environmental parameter estimation sliding mode，EPESM）。以阀控液压缸的活塞位移传递函数为基础建立了基于位置内环的SEA阻抗控制模型，并以PID作为基础控制器；为改善SEA阻抗控制的动态性能，根据Lyapunov第二法构建稳定的自适应环境参数估计方法对SEA期望位置进行前馈补偿；为提升自适应环境参数估计方法在SEA工作过程中不同阶段的动态性能和环境变化适应性，使用模糊控制方法对自适应环境参数估计方法中的自适应参数进行寻优；以SEA状态方程为基础构建滑模控制器与PID控制器进行动态性能对比分析。仿真结果表明：在变SEA弹簧刚度工况和变环境刚度下，EPESM阻抗控制的响应速度明显更快，可将调节时间从平均5 s缩短到1 s内，能更快地达到预期位移和预期接触力，且能略微降低稳态误差，使接触力误差保持在±6 N内。在动态跟踪工况下，EPESM阻抗控制的动态性能更好，在快速进入跟踪状态后，可以长时间保持0.2 s以内的相位滞后和5.2%的幅值误差。

Heavy quadruped robots are subjected to uncertain impact loads during foot-to-ground contact and gait transition, which can easily lead to excessive load on the foot mechanism and structural impact damage. Therefore, a sliding mode impedance control method based on environmental parameter estimation (EPESM) was proposed to solve the problem of poor dynamic performance when using hydraulic series elastic actuators (SEA) as foot ends in unstructured environments. Based on the piston displacement transfer function of the valve controlled hydraulic cylinder, an SEA impedance control model based on the position inner loop is established, with PID serving as the basic controller. To improve the dynamic performance of SEA impedance control, a stable adaptive environment parameter estimation method based on Lyapunov’s second method is constructed to compensate for the expected SEA position using feed forward compensation. To improve the dynamic performance and adaptability of adaptive environmental parameter estimation methods at different stages of SEA work, fuzzy control methods are used to optimize the adaptive parameters in these methods. Based on the SEA state equation, a sliding mode controller and a PID controller are constructed for dynamic performance comparison and analysis. Simulation results show that under variable SEA spring stiffness and variable ambient stiffness conditions, the response speed of EPESM impedance control is significantly faster. The adjustment time can be significantly reduced from an average of 5 s to within 1 s, achieving faster expected displacement and expected contact force, while keeping the steady-state error slightly reduced and the contact force error within ±6 N. Under dynamic tracking conditions, EPESM impedance control exhibits better dynamic performance, maintaining a phase delay within 0.2 s and an amplitude error of 5.2 % for an extended period after quickly entering the tracking state.

TP242

国家自然科学基金资助项目(51509006)；先进节能教育部工程中心开放课题(SWEDT-KF201902)。