Digital Optimal Control of Rigid Manipulators

摘要

The thesis demonstrates that solving digital control problems without making any approximations allows for the design and use of digital robot controllers with 'large' sampling times. The determination of the desired robot motion as a function of time often involves a nonlinear optimal control problem. This problem is generally solved assuming a continuous-time control. The author demonstrates that explicitly using the piecewise constant nature of the digital control may simplify the solution of this nonlinear optimal control problem. He focuses on three types of robot motion control problems where he considers both rigid cartesian robots, having basically linear dynamics, and other types, having non-linear dynamics. He solves the time-optimal control problem where the initial and final positions and velocities of the robot links are prescribed and the objective is to realize the transition in minimum time, given bounds on the control variables. He solves the time-optimal path tracking problem where the link positions are prescribed as a function of a certain parameter. Finally, he solves the tracking problem where the link positions are specified as a function of time.