Abstract: In the process of robot human-machine interaction, stable collision detection is the guarantee of human-machine safety and the safety of the equipment itself. Therefore, this paper proposes a robot collision detection method with load and no additional sensors. Through the compensation of the robot dynamics model, the error of the motor torque of each joint of the robot is analyzed, and the load identification and collision detection are realized. In this paper, various parameters of the load are added to the dynamic calculations to compensate for the errors caused by the load on the dynamics calculations. Through the analysis of the torque error of each joint motor, the error extreme value in each speed segment is screened, and the envelope of the upper and lower limits of the error-speed curve is fitted, and the dynamic threshold is set. The simulation results show that the established dynamic compensation model can effectively identify the load, eliminate the interference of the end load, and can accurately realize the collision detection function with the change of the robot's movement speed.

Key words: Collision detection; Load compensation; Motor torque error; Dynamic threshold; Robot

摘要： 在人机交互过程中，稳定的碰撞检测是人机安全和设备自身安全的保障，因此提出了 一种有负载、无附加传感器的机器人碰撞检测方法。通过对机器人动力学模型的补偿，分析 机器人各关节电机力矩的误差，实现负载识别与碰撞检测。将负载的各项参数添加到动力学 计算中，补偿因负载而对动力学计算造成的误差。通过对各关节电机力矩误差的分析，筛选 每个速度分段内的误差极值，并对误差-速度曲线上下限的包络线进行拟合，设置动态阈值。 仿真结果表明，建立的动力学补偿模型能够有效识别负载，消除末端负载的干扰，并且能够 随着机器人运动速度的变化准确地实现碰撞检测功能。

关键词: 碰撞检测；负载补偿；电机力矩误差；动态阈值；机器人