基于六维力传感器的机器人末端执行器碰撞位置检测方法

Collision position detection for robotic end-effector using force/torque sensor

末端执行器的碰撞位置检测是机器人力觉感知的重要组成部分,也是机器人与外界环境交互不可或缺的功能。仅依靠碰撞力和力矩数据求解碰撞位置的难点在于多解问题,需要借助额外的约束条件来求解唯一解。现有的方法大多采用几何约束条件,其缺点在于无法应用于非结构化的环境和无法应对碰撞物体的形变。为了弥补这一缺陷,本文建立了一种动态力学数据自约束碰撞定位模型,依靠时间轴上的多组力和力矩数据互为约束条件,并建立了最小误差搜索策略来求得最优解。与传统方法相比,改进算法不需要再依赖于碰撞物体表面的几何信息,大大增强了其实用性,扩展了应用范围,并提高了位置检测的精度和鲁棒性。实验结果证明,改进的算法在碰撞表面几何模型未知,或发生变形时均发挥作用,优于传统方法。

Collision position detection of the end-effetor is an important part of force sensing,which can help the robot interact with the environment.The difficulty of calculating the collision position using force and torque data is multi-solution problem,and external constraints need to be introduced to find the unique solution.The previous methods all depend on the geometric constrain,and unable to cope with unstructured environments and deformation of the end-effector.To remedy the defects of the previous methods,a self-constrained collision localization model based on dynamic force and torque data was proposed,which had got rid of the geometric constraints and enhanced the robustness with the help of minimum-error method.Therefore,the application range was expended greatly.The experimental results showed that the proposed method was better than the traditional method.