外骨骼机器人人体姿态解算算法研究

Study on the human posture algorithm of exoskeleton robot

摘要:

外骨骼机器人的人机协同控制因其自身特点,具有较大难度。目前,虚拟力矩控制是一种较为常见的控制方式,但其基于物理特性以及运动学参数预测人机作用力的原理决定了该方法必须依靠极其精确的惯性系参数预测才能实现(假设系统关节处无摩擦),大量的系统动力学方程运算也给控制器带来了较大负担。针对该问题提出了一种基于位置随动的控制算法,通过在人机接触点直接引入激光流式间距传感器测量人机(外骨骼)间距信息,结合外骨骼姿态信息计算由人体各关节角度构成的人体姿态,作为外骨骼随动的目标值,进而计算人体末端位置,最终解算出能达到同样末端位置的外骨骼姿态。该方法具有运算量下、稳定性好、控制精度高等特点。

Abstract:

Virtual-force control is based on estimating human-machine forces through mass properties and kinematic parameters. A drawback of this method is that it relies on the accuracy of the inertial estimation, assumes a system with frictionless joints, and that the dynamic system equations require a large amount of computations. The introduction of laser flow spacing sensor between the human and the exoskeleton can solve some of these issues. This paper proposes a position servo control algorithm, by introducing the laser directly in the man-machine contact streaming man-machine information (exoskeleton) spacing distance sensor measurement, combined with the exoskeleton posture information calculated by the human body each joint Angle in the composition of the human body posture, as the target of exoskeletons follow-up, terminal position and then calculate the human body, calculated finally able to achieve the same position at the end of the exoskeleton. This method has the characteristics of high performance, stable stability and high control precision.