基于改进动态运动基元的6D轨迹规划

动态运动基元(DMPs)轨迹规划方法可以简化机械臂控制中参数调整的复杂过程,快速生成运动轨迹,但是面对姿态的流形特性以及跨零点情况,现有的DMPs很难达到预期的效果.本文提出了一种基于改进DMPs的笛卡尔空间6D轨迹规划方法.该方法采用四元数描述姿态,实现了位置轨迹与姿态轨迹的无奇异表示.通过解耦强迫函数与起–终点状态差值项之间的关联,消除了跨零点引起的轨迹抖动、无法生成与翻转等问题.此外,基于机械臂和障碍物间的距离与偏角建立了虚拟阻抗关系,并将其耦合到动力学模型中,实现了机械臂末端的避障控制,避免了避障行为过早问题,有利于减少消耗.机械臂6D轨迹规划仿真和实验表明,本文提出的改进DMPs方法有效.

6D trajectories planning based on improved dynamic movement primitives

The dynamic movement primitives (DMPs) trajectories planning method can simplify the complex process of parameter adjustment in the robotic arm control and quickly generate the movement trajectories. However, in the face of the manifold characteristics of the orientation and the zero-crossing situation, the existing DMPs are difficult to achieve the expected results. This paper proposes a 6D trajectories planning method in Cartesian space based on improved DMPs. This method uses quaternion to describe the orientation, and realizes the non-singular representation of the position trajectories and the orientation trajectories. By decoupling the correlation between the forcing function and the difference term of initial-target state, the problems of trajectories jitter, inability to generate and flip caused by the zero-crossing point are eliminated. In addition, based on the distance and deflection angle between the robotic arm and the obstacle, the virtual impedance is established and coupled to the dynamic model. The obstacle avoidance control at the end of the robotic arm is carried out to avoid the problem of premature obstacle avoidance and reduce consumption. The simulation and experiment of robot arm 6D trajectories planning show that the improved DMPs method proposed in this paper is effective.