Research On Real-time Control System For The Baby Elephant

Robot which works as the extensive sensor for human beings hasbeen utilized more and more widely in industry and human life. Thewheeled or tracked vehicles can only carry out tasks within a limitedarea, which not sufficient to meet the needs of human beings. While thewalking robot, which is of a capability to adapt to a wider range ofterrains, will expand the range of activities for humans. For example, itcan be well applied in nuclear power plants, coal mining wells and spaceexploring. Thus, it is very important to carry out the research anddevelopment on walking robots.In the unstructured or dangerous environment, a walking robot needsto respond to the uncertain environment timely, for example, therecovering from stepping on obstacles unexpectedly or losing its gait. Therobot would have fallen if the control system takes too much time tofeedback. Therefore, real-time performance is the foundation for awalking robot to adapt to the environment. This thesis studies thereal-time performance based on the Baby Elephant, and the main work ofthis paper is described as the following:1．Participate in the design and manufacturing of the mechanicalstructure for the Baby Elephant, and design the mechanism for thequadruped robot, and provide the robot kinematics solutions. Afterdesigning the quadruped robot, we develop the prototype of the robot.2．Develop an open-architecture control system based on hostcomputer and motion control card. Analyze the requirements of walking robot control system. The host computer is PC104, which mainlyresponses for human-computer interaction, decision-making andtrajectory planning and also signal processing, the lower computer isTurbo PMAC to implement servo control for the robot, with theutilization of Elmo amplifiers and Maxon motors. The control hardwaresolution is superior than most of traditional control systems, and is ofpowerful information processing capability, very well scalability andportability.3． Configure the servo control parameters and implement asmooth-speed trajectory interpolation algorithm. After building up themotor model and analyzing the servo control algorithms of both PMACand Elmo driver, I optimize the PID parameters, and set servo frequencyand other related parameters. The newly method to calculate the endstatus of each adjacent PVT pieces has been derived, which can achievethe continuous trajectory with smooth velocity on the whole.4．Design and develop a real-time control software. Based on LinuxRTAI real-time operation system, and improve the real-time performanceof the PMAC driver, design and code the core threads as well to achievemotor control.5．Develop the platform and test the performance of the real-timecontrol system. Develop a real-time control hardware and softwareplatform. Test the real-time performance for the operation system,real-time performance for real-time driver and overall systemperformance for the motor control. The experimental results show that thesystem performs well both on trajectory interpolation and guaranteedtime in response, which can meet the requirements of the control of BabyElephant.