Search for: All records

We propose a new concept named subschedulability to relax schedulability conditions on task sets in the context of scheduling and control co-design. Subschedulability is less conservative compared to schedulablity requirement with respect to network utilization. But it can still guarantee that all tasks can be executed before or within a bounded time interval after their deadlines. Based on the subschedulability concept, we derive an analytical timing model to check the sub-schedulability and perform online prediction of time-delays caused by real-time scheduling. A modified event-triggered contention-resolving MPC is presented to co-design the scheduling and control for the sub-schedulable control tasks. Simulation results are demonstrated to show the effectiveness of the proposed method. 

We analyze a human and multi-robot collaboration system and propose a method to optimally schedule the human attention when a human operator receives collaboration requests from multiple robots at the same time. We formulate the human attention scheduling problem as a binary optimization problem which aims to maximize the overall performance among all the robots, under the constraint that a human has limited attention capacity. We first present the optimal schedule for the human to determine when to collaborate with a robot if there is no contention occurring among robots' collaboration requests. For the moments when contentions occur, we present a contention-resolving Model Predictive Control (MPC) method to dynamically schedule the human attention and determine which robot the human should collaborate with first. The optimal schedule can then be determined using a sampling based approach. The effectiveness of the proposed method is validated through simulation that shows improvements. 

Expert based ensemble learning algorithms often serve as online learning algorithms for an unknown, possibly time‐varying, probability distribution. Their simplicity allows flexibility in design choices, leading to variations that balance adaptiveness and consistency. This article provides an analytical framework to quantify the adaptiveness and consistency of expert based ensemble learning algorithms. With properly selected states, the algorithms are modeled as a Markov chains. Then quantitative metrics of adaptiveness and consistency can be calculated through mathematical formulas, other than relying on numerical simulations. Results are derived for several popular ensemble learning algorithms. Success of the method has also been demonstrated in both simulation and experimental results.