Neural Event-Triggered Control with Optimal Scheduling

Luan Yang, Jingdong Zhang, Qunxi Zhu, Wei Lin

TL;DR: We design a controller that only sends updates when absolutely necessary, without sacrificing safety or stabilization performance.

Abstract: Learning-enabled controllers with stability certificate functions have demonstrated impressive empirical performance in addressing control problems in recent years. Nevertheless, directly deploying the neural controllers onto actual digital platforms requires impractically excessive communication resources due to a continuously updating demand from the closed-loop feedback controller. We introduce a framework aimed at learning the event-triggered controller (ETC) with optimal scheduling, i.e., minimal triggering times, to address this challenge in resource-constrained scenarios. Our proposed framework, denoted by Neural ETC, includes two practical algorithms: the path integral algorithm based on directly simulating the event-triggered dynamics, and the Monte Carlo algorithm derived from new theoretical results regarding lower bound of inter-event time. Furthermore, we propose a projection operation with an analytical expression that ensures theoretical stability and schedule optimality for Neural ETC. Compared to the conventional neural controllers, our empirical results show that the Neural ETC significantly reduces the required communication resources while enhancing the control performance in constrained communication resources scenarios.

Lay Summary: Modern controllers powered by machine learning can make smart decisions in complex systems—like self-driving cars or robotic arms—but they often demand constant updates and heavy data communication to stay stable and effective. This makes them hard to use in real-world devices where communication is limited, like drones or satellites. We tackled this by designing a controller that only sends updates when absolutely necessary, without sacrificing safety or performance. Our new method, called Neural ETC, learns when to act and when to stay silent, reducing the number of communications needed. The result: smarter, quieter controllers that work better in the real world. This brings us one step closer to deploying intelligent systems in places where every bit of data—and every second of response time—counts.

Application-Driven Machine Learning: This submission is on Application-Driven Machine Learning.

Link To Code: https://github.com/jingddong-zhang/Neural-Event-triggered-Control

Primary Area: Optimization->Everything Else

Keywords: Event-triggered control, Stability guarantee, Limit resources, Optimal scheduling

Submission Number: 6617