Design of an integrated environment for operation and control of robotic arms (non-reviewed)

Abstract

As more advanced control algorithms are becoming available for the control of robotic arms, traditional fixed controller boards and associated code generators are becoming less convenient way to test such control algorithms in real-time. The process of using such boards is complex, time consuming, and inflexible. In this work, an integrated hardware-software environment was developed and presented where researchers can simply use any Matlab/Simulink basic function block and/or toolbox, such as fuzzy logic or neural network, to design, implement, and test different controller algorithms in realtime for robotic arm operations. The hardware includes a computer, the dSPACE-ds1103 digital processing board, an amplifier board, and the Zebra-ZERO robotics arm as a test-bed. Also, Matlab GUI, m-file, Matlab/Simulink blocks, and dSPACE interface functions are combined together to form the software environment. Control algorithms can be designed in the Matlab/Simulink then converted to c-code and download to the dSPACE processing board. The Matlab m-file are used to code the arm inverse kinematics model and the path planning to calculate the joint angles then send them to the dSPACE processing board using the dSPACE interface functions. Finally, the dSPACE processing board generates physical signal to control the robot arm in real-time. The proposed hardware-software components are developed and integrated together, and several control algorithms can be tested on it. The development steps and some of the realtime testing results conducted on the hardware are explained next in this extended abstract. Typically, controllers are designed to run on dedicated hardware and researchers need different hardware to test different control strategies. This can be costly and time consuming where one has to develop different control environment for every control strategy to be tested. In this work, an integrated hardware-software environment was developed for implementation and testing of different control algorithms in real-time. The integrated system is composed of a computer, a power supply, the DS1103 dSPACE controller board, an amplifier, and the Zebra- Zero force robotics arm. The computer is used to send commands to the DS1103 dSPACE controller board.Inside the DS1103 dSPACE controller board, a Texas instruments DSP micro-controller performs the necessary calculation to determine the PWM signal to be generated and sent to the amplifier. The amplifier then generates the control signals that are applied to dc-motors that drive the links. The motor encoders provide feedback position signals as output. To develop the software environment, the Matlab programming environment (m-file), Matlab's graphical user interface, Simulink, and the toolbox are all employed. A user graphical interface (GUI) was designed for user convenience. The robot can be moved to the ready position then, the forward or inverse kinematical model is chosen according to the type of input data. The links begin to move when the Move button is pressed. The user can also select different movement speed for each link. Finally, when link movement has ceased, the joint trajectories are displayed on the GUI. Trajectory planning files for position, velocity and acceleration references are also developed and implemented in the environment. Two types of trajectories are made available according to different requirements; second order polynomial and third-order polynomial trajectory. The second order polynomial trajectory is recommended for links with large angular position difference. For purpose of testing and verification, the Zebra-Zero robotics arm was used. The Lagrangian mechanics is used to develop the dynamic equations for the Zebra-Zero robotic arm. Some of the arm parameters are calculated while others are determined experimentally, e.g., the link inertias and masses. A Simulink model of the robotic arm dynamic was developed. To test the environment a control algorithm was also designed then automatically converted to C programming language and downloaded to the DS1103 dSPACE controller board. The user enters commands using the Matlab GUI. Based on input, positions or final location and orientation, the forward or inverse kinematical model is selected. In this work a PID control algorithm was designed and tested on the Zebra- Zero robotics arm. To verify the controller performance, Matlab toolbox was used to simulate the Zebra-Zero robotic arm dynamics model. The results were very comparable with the actual Zebra-Zero robotic arm hardware performance.