FPGA based hardware accelerator for parallel robot kinematic calculations

Abstract

The paper presents an application of FPGA to support the calculation of the inverse kinematics problem of a parallel robot. The presented robot is designed for milling by moving the spindle along a desired trajectory generated in Cartesian space. This means that for each point of the trajectory solution of the inverse kinematics problem is needed. The resulting sequence of data creates the joint space trajectory. The trajectory in joint space must be calculated in real time. Required high frequency and complex equations makes the problem of the calculation time crucial. The paper shows how to increase the computing power for inverse kinematics problem solving, preserving required calculation accuracy, by augmenting the arithmetic coprocessor with custom instructions. The paper shows hardware implementation of the accelerator and presents results of calculations per-formed on Altera FPGA chip.