![]() | Only 14 pages are availabe for public view |
Abstract As robots are being used in a growing range of applications, the issue of reliability and fault tolerance becomes very important. But, in the presence of actuator failures, the robot is an underactuated system with less actuators than the number of joints and this underactuation introduces second order non integrable constraints that relate active joint accelerations to passive joint accelerations. This thesis studies the problem of real time path planning for robots in the presence of actuator failures, which is the problem of specifying the joint space path for the execution of the end-cffcclor cartesian space path. The real time path planning problem is posed as a finite time nonlinear control problem which is solved in parallel with an optimization technique in the real time. To include the second order constraints introduced by the presence of actuator failures, we constructed an augmented system using the second order cartesian space dynamic equations. This augmented system control problem is then solved to get the joint accelerations thai guarantee asymptotic stability of the augmented system, i.e., tracking the desired trajectory and, at the same time, satisfying the second order constraints. For the solution of this control problem, we developed a new controller that we called the Fault Tolerant Motion controller (FTM) . Numerical integration was used to get an estimate 7 for the zero order and first .order error vectors. A control input is added to compensate for the bounded error introduced by this numerical integration. Then, we presented a proof of asymptotic stability for the FTM control problem using the second method of Lyapunov for nonlinear systems. Based on the effect that the Jacobian of the augmented system has on the performance of the FTM, we proposed a new measure that we called the Reliability Index (RI). This index can be used to monitor, instantly, the performance of the FTM. Also, the use of this index in some other problems for underactuated manipulators is illustrated. These problems include: the optimal actuator placement problem, the optimal cartesian coordinate problem, the problem of optimal choice of the robot parameters, and the set point kinematic control problem. |