| This doctoral dissertation describes real-time software systems for use in control and robotic applications. The body of this dissertation focuses on three systems developed at Clemson University: (i) QMotor 2.0, a single-processor based system for implementing and tuning control programs in C, (ii) Motor 3.0, an object-oriented system for implementing and tuning control programs in C++, and (iii) QRobot, a single-processor based control system for Puma robotic manipulators.; Traditionally, digital control programs were implemented using a heterogeneous multi-processor architecture, typically consisting of a host general purpose computer (PC, Mac, SGI, etc.) and an embedded digital signal processor (DSP) based single board computer (SBC). This architecture was required because general purpose computers (GPCs) either did not have enough processing power to compute a control law at high frequencies, and/or did not run real-time operating systems that would allow them to execute the control programs deterministically. The first chapter describes QMotor 2.0, a system that integrates modern PC CPUs and a real-time operating system to produce a uniprocessor architecture with greater functionality and lower cost and complexity than the traditional multi-processor host/ DSP architecture.; The second chapter describes QMotor 3.0, the successor to QMotor 2.0. As QMotor 2.0 was used constantly at Clemson University, many shortcomings became apparent. The field of Control Engineering requires great flexibility in implementing experiments. New hardware interfaces must constantly be added to take advantage of the latest sensors and actuators. The functional programming design of QMotor 2.0 limited the ability to re-use code, and to easily modify and maintain the code. QMotor 3.0 is a complete redesign of QMotor 2.0, using object-oriented programming (OOP) principles throughout, and implemented in the C++ programming language.; The third chapter describes QRobot, a robot control system that builds on the concepts and technology of QMotor 3.0. Traditionally robot control systems are composed of closed, multiprocessor architectures. The closed architecture ensures that users do not have access to the control algorithms used to drive the robot. This prevents users from implementing more sophisticated control algorithms that become possible as CPUs become more powerful. It also prevents the integration of new sensors (such as force/torque sensors and video cameras) into the control loop. QRobot demonstrates how a closed, proprietary, firmware-based robot control system can be replaced with an open, PC-based control system, lowering cost and complexity, and increasing flexibility and power. |
