Search for: All records

We report our experiences implementing standards-based grading at scale in an Algorithms course, which serves as the terminal required CS Theory course in our department's undergraduate curriculum. The course had 200-400 students, taught by two instructors, eight graduate teaching assistants, and supported by two additional graders and several undergraduate course assistants. We highlight the role of standards-based grading (SBG) in supporting our students during the COVID-19 pandemic. We conclude by detailing the successes and adjustments we would make to the course structure. 

Deterministic and stochastic approaches to handle uncertainties may incur very different complexities in computation and memory, in addition to different uncertainty models. For linear systems with delay and rate constrained communications between the observer and controller, previous work shows that the deterministic approach l_infty control has low complexity but only handles bounded disturbance. In this paper, we take a stochastic approach and propose an LQ controller that can handle arbitrarily large disturbance but has large complexity in time/space. The differences in robustness and complexity of the l_infty and LQ controllers motivate the design of a hybrid controller that interpolates between the two: The l_infty controller is applied when the disturbance is not too large (normal mode) and the LQ controller is resorted to otherwise (acute mode). We characterize the switching behavior between the normal and acute modes. Using theoretical bounds and supplementary numerical experiments, we show that the hybrid controller can achieve a sweet spot in robustness-complexity tradeoff, ie, reject occasional large disturbance while operating with low complexity most of the time. 

We frame the collision avoidance problem of multi-agent autonomous vehicle systems into an online convex optimization problem of minimizing certain aggregate cost over the time horizon. We then propose a distributed real-time collision avoidance algorithm based on the online gradient algorithm for solving the resulting online convex optimization problem. We characterize the performance of the algorithm with respect to a static offline optimization, and show that, by choosing proper stepsizes, the upper bound on the performance gap scales sublinearly in time. The numerical experiment shows that the proposed algorithm can achieve better collision avoidance performance than the existing Optimal Reciprocal Collision Avoidance (ORCA) algorithm, due to less aggressive velocity updates that can better prevent the collision in the long run.