Sensing Information Forecasting Algorithm And Controller Design For Wearable Power Assist Robot

The Wearable Power Assist Robot is a kind of assistive rehabilitation robot. It is an autonomous exoskeleton which is to help people to expand the capacity of the lower limbs motion. In order to acquire the intention of human body movement, the sensing system obtains the sensing information from human body, based on this sensing information, and then the control system takes charge of the DC servo motors which are placed on the hip joints and knee joints to provide the assistive power. The sensing system of the Wearable Power Assist Robot is built up with the two-dimensional force sensors, the one-dimensional force sensors and the angle sensors. The goal of using the force sensors is to acquire the intention of human body movements by means of detecting force. The two-dimensional force sensors are placed on the thigh link and the lower thigh link, and the one-dimensional force sensors are placed in the footboard, and the encoders in the DC motors are used as angle sensors. In practice, whether the sensing system can acquire human motion intention from human body accurately and rapidly is becoming the key issue in our research. In order to improve the dynamic response of the exoskeleton and ensure the real-time quality of the whole system, we propose a novel sensing information forecasting algorithm which is based on the Time Series Analysis. This forecasting algorithm can be used on-line and increase the dynamic response of the sensing system many times. In order to achieve the control system of small size, light weight and good scalability, we also design and produce the first generation of motion controller for the DC servo motor which is utilized in the Wearable Power Assist Robot, this motion controller mainly utilizes LM629 which is a precision motion controller chip and LMD18200 which is a H-bridge driver chip for motion control applications. We can simplify the system design and increase the system integration by using them.This paper is supported by the National Science Foundation of China (Grant #60575054) and Foundation of 863 Program (Grant #2006AA040204).The main contents of the paper are as follows:1. The design of the sensing forecasting algorithm based on the time series analysis: In order to meet the requirement of small amount of computation, high precision and can be used on-line, we build up a novel sensing information forecasting algorithm, the algorithm is made up with the autoregressive model, the recursive least square method and the final prediction error order selection criterion.2. The design of software and hardware system for the sensing information forecasting algorithm: In order to meet the requirement of real-time and high speed, we design the software and hardware system for the sensing information forecasting algorithm. The hardware is made up with the signal conditioning circuit and the signal acquisition, processing and transmission circuit. The software can be categorized into the software in MCU and the software in PC, and the software in PC is achieved by the VC++ and MATLAB mixed programming.3. The simulations and experiments of the sensing information forecasting algorithm: Firstly, we utilize MATLAB to calculate the results of the algorithm, and then, we can compare the predictive value with the measured value to verify the validity of the algorithm and derive the relationship between the forecasting error and the forecasting steps. Finally, the correlative experiments have been carried out, and the results fully demonstrate the effectiveness and practicality of the sensing information forecasting algorithm.4. The design of motion controller for the Wearable Power Assist Robot: In order to meet the requirement of small size, light weight, high integration and good expansion, we design the first generation of the motion controller. This controller uses the embedded system to achieve a high level of integration. Through the serial port, the controller can be debugged and achieve communication. The motion controller lays the foundation for the future whole embedded control system design.