Adaptive and Delay-Compensating Robot Controllers

The robot manipulators are used in network-based industrial units, and even homes, by expending a significant lumped amount of energy. Manipulators are also subject to large time delays like many engineering systems, which may lead to the lack of high precision required and could cause not only poor control performance, but also catastrophic instability. Hence designing control algorithms for such delay systems to compensate for the time delay and having a robust control law is a necessity. Another important point which needs to be noticed in utilizing robots is environmental uncertainties, which may not be considered in the modeling. To summarize, parametric uncertainty and/or time-delay through manipulation along with optimal operation are important criteria for the operation of robotic systems. One of the main contributions of this dissertation is focusing on the development of robust techniques for achieving optimal operation and control of the arm to attain perfect tracking. For the operational optimization, we examine a discrete-time multi-variable gradient-based Extremum Seeking (ES) scheme enforcing operational time and torque saturation constraints to minimize the lumped amount of energy consumed for a path given. Finally, the optimal trajectory is experimentally implemented to be thoroughly compared with the inefficient one. Precise control of manipulators in the presence of delay or uncertainty and variation in their environments is also a prerequisite to feasibly utilize robot manipulators. Nonlinear control theorems for nonlinear systems (e.g. robotic systems) have been developed for a long period, however, implementing them, in reality, is one of the most challenging problems in engineering applications. Therefore, formulating novel robust and computationally efficient control approaches is still a necessity. Another significant contribution of this work is a practical implementation of various control theories, with verifying all the necessary assumptions, for high-DOF manipulators. We formulate different computationally efficient control laws to implement real-time controllers for a high Degree-of-Freedom manipulator with uncertainties or in the presence of delay to move toward implementing those control laws in other engineering applications.