Delay-Aware Control for Autonomous Systems

The increasing demand for implementing multiple perception-based real-time applications (e.g., lateral control) onto autonomous systems is introducing new challenges into their control design paradigm, such as the delay-aware design of software controllers for such systems. This becomes more challenging when a variable delay occurs due to the variation of the workload of computation associated with the perception tasks, such as compute-intensive image processing tasks. For such systems, the traditional controller-design-followed-by-implementation paradigm fails to provide the desired control performance all the time. One such example is designing controllers by considering worst-case perception computation delay and thus providing sub-optimal control performance. The primary focus of this work is to address this issue by introducing a novel multi-rate control synthesis approach for autonomous systems. The idea is to apply multiple controllers and adaptively switch among them based on the current delay value without compromising the stability while improving the control performance. We report promising results w.r.t state-of-the-art approaches toward enhancing control performance and optimizing control cost.