一种参数自适应的机器人碰撞检测方法，首先对机器人进行动力学建模及动力学参数辨识，再设计基于广义动量的扰动观测器，用于机器人外部碰撞力的检测，然后设计机器人动力学参数自适应方法，最后设置动态变化的动量观测器残差阈值，采集机器人运行过程中的角度、角速度以及电流信号作为算法输入，当估计的外部碰撞力超过阈值范围时，判断为机器人与外界物体或操作人员发生碰撞；本发明简单有效，不需要增加额外的传感器，且能够有效解决机器人动力学参数辨识不准确且在运行过程中动力学参数发生变化的问题，提高碰撞检测的准确性。

A parameter adaptive robotic collision detection method, first performing dynamic modeling and dynamic parameter recognition of the robot, redesigning a generalized momentum-based perturbation observer, for detection of robot external crash forces, a robot dynamic parameter adaptation method is then designed and finally set dynamically changing momentum observer residual thresholds, acquiring angle, angular velocity and current signals during robot operation as algorithm inputs, and when the estimated external collision force exceeds the threshold range, the robot is judged to collide with an external object or human operator; the present invention is simple and effective, does not require additional sensors to be added, and can effectively solve the problem of inaccurate recognition of dynamic parameters of a robot and changes in dynamic parameters during operation, improving the accuracy of collision detection.