Real-time Control Modelling and Output Signal Data Analysis Based on the Stewart Platform

Abstract: Around 40 million people around the world have amputated parts of their bodies. Prostheses are widely used to support their daily behavior. However, about 35% of amputees refuse to use the prostheses due to comfort problems. SocketSense is an EU research project that aims to improve the situation. The Stewart platform (SP) is used as the model of the prototype. The SP could reproduce the comparing forces and torques interior the socket-to-stump interface, given prerecorded movement information of an amputee’s walk cycle. However, previous work controls SP based on inputting some simple reference trajectories ,or for radar, satellite and vehicle use-cases. This  thesis is committed to designing and implementing a control algorithm for the SP to control the socket. The task is divided into two main parts: simulation on Matlab/Simulink and implementation on Linux/ROS, and each part is split into several steps. Firstly, the PID control algorithm is selected and developed for the SP. Secondly, the simulation model is designed and built on Matlab/Simulink. Based on the simulation model, the control algorithm is implemented into the model. Then, the results of simulation are obtained. After that, the SP model is built on Linux/ROS. Then, the model is connected and communicated with the real platform. Finally, the output data and system performance are analysed in terms of control performance, real-time performance, anti-noise performance, etc. The results show that both the simulation and the actual platform system can follow the reference trajectory well. Although there is noise in the results of the actual platform, the noise is suppressed to a great extent by Kalman filter.