目的 针对存在系统未建模特性和负载变化下码垛机器人关节空间轨迹跟踪控制的问题，设计一种基于结合时延估计技术与自适应积分滑模面的控制策略。方法 根据圆饼工件分拣需求，设计一款桌面式码垛机器人系统，推导机器人的运动学与动力学模型，给出关节空间轨迹规划算法，并基于无模型思想设计关节空间轨迹跟踪控制器。结果 利用雅克比伪逆法可反解出机器人的关节角;通过所提的轨迹规划算法能有效获得各关节运动轨迹;与PID控制器和积分滑模控制器相比，文中所提控制器具有较好的控制精度、较强抗干扰性和较高的鲁棒性。结论 仿真和实验结果表明，所设计的基于时延估计技术的自适应积分滑模控制器是合理的，能使得码垛机器人完成圆饼工件的分拣任务，具有一定的工程应用价值。

The work aims to propose a control strategy combing time delay estimation technique and adaptive integral sliding mode for trajectory tracking control in joint space of palletizing robot under unmodeled characteristics and load variation. According to the sorting requirements of round pieces, a desktop palletizing robot system was designed. The kinematics and dynamics models of the robot were derived to obtain the trajectory planning algorithm in joint space. Then, the trajectory tracking controller in joint space was designed based on the model-free idea. The joint angles of the robot were obtained by the Jacobi pseudo-inverse method and the trajectory of each joint was acquired through the proposed trajectory planning algorithm. Compared with PID controller and integral sliding mode controller, the proposed controller had better control accuracy, stronger disturbance rejection and higher robustness. The simulation and experiment results prove that the adaptive integral sliding mode controller designed based on time delay estimation technique is reasonable and can help the palletizing robot deal with the tasks of round pieces sorting, which has a certain engineering value.