Vision-Based Paddle Juggling System with Carpal Wrist Robot

Abstract

The main objective of this thesis is to design and implement a vision-based paddle juggling system using a stereo vision system and a carpal wrist robot The carpal wrist robot is chosen because of its superior dexterity and similarity to the human wrist and it has three degree of freedom As a result of its symmetric parallel architecture it has a large payload capacity In the stereo vision system two image sensors are used to provide stereo vision The stereo vision system is able to track the ball according to its color information An extended Kalman filter is then used to estimate the dynamics of the ball and predict its future trajectory and velocity Based on the dynamics of the ball estimated by the extended Kalman filter the carpal wrist robot can determine the timing velocity position and posture for paddling the ball Image processing algorithms are implemented on an FPGA-based board through Verilog hardware description languages The extended Kalman filter and robot controller are implemented on digital signal processors Through experiments the ball can be tracked by the stereo vision system and the carpal wrist robot can continue to paddle the ball and the ball is maintained at set height with at least 150 successful strokes

Date of Award	2019
Original language	English
Supervisor	Ming-Tzu Ho (Supervisor)