Intelligent Robot Arm Pick and Place using Hough transform vision algorithm

Abstract

: This paper proposes an improved algorithm for 4-degree of freedom robot arm forward/inverse kinematics with vision sensor fusion utilizing Hough transform for a pick and place application. As a result, an optimized approach for path planning and end-effector velocity generation is enabled. The main advantage of the algorithm is the simplicity of deriving the manipulator kinematic equations and generating solutions. This is achieved by deriving a common set of simultaneous equations that characterize the system for both forward/inverse kinematics, and subsequently solving numerically using Matlab symbolic toolbox. The solution incorporates input from vision sensor for object detection and tracking by implementing OpenCV vision libraries under Matlab, interfaced with a low cost microcontroller that controls the manipulator servos. The path planning approach involves segmentation of identified optimal path and stabilizing end-effector velocity by calculating corresponding joint angular velocity of each segment. To test the effectiveness of the proposed algorithm, Matlab simulations are conducted and preliminary results are produced for three components namely the kinematics algorithm, vision Hough Transform besides the optimal path planning approach. The superiority of this proposal is in the simplicity, compared to the widely used homogeneous matrix transformation that implements the cumbersome Denavit-Hartenberg convention for setting up the frame coordinate, also the proposed algorithm removes singularities and redundancies from the solution and confine the solution to the servos operating limits. It is envisioned that this algorithm will be efficient in the design of practical high performance, low cost robot arm in preparation for the ongoing industrial revolution.