Fault-tolerant Predictive Control with trajectory planning in the presence of obstacles

Abstract

Predictive Control techniques have been increasingly widespread in industry, mainly due to their ability to handle constraints on inputs and states of the plant. Approaches equipped with terminal constraint invariant sets are frequently used to achieve stability of the control loop. However, in vehicle guiding missions, the terminal set can be imposed by mission requirements and not necessarily invariant. Another difficulty is the occurrence of faults that limit the authority of the actuators, compromising the ability to achieve the specified terminal set within the horizon, thus requiring the use of larger horizons. Moreover, the presence of obstacles makes the optimization problem in general non-convex. These factors contribute to increase the computational load associated with the solution of the optimization problem that generates the control signal. Therefore, the time required for solution can be prohibitive in front of the sampling period. In this context, it is convenient to have a trajectory planner capable of dividing the task into parts, allowing the use of smaller horizons. In this work, path planning techniques in the presence of obstacles and dynamic constraints will be studied, which define waypoints, i. e., intermediate points, to be used by the predictive control loop. The results will be analyzed in terms of demanded computational resources, as well as the ability to meet mission requirements even in the presence of actuator faults.