MeanFlow Meets Control: Scaling Sampled-Data Control for Swarms
2026-03-20 • Machine Learning
Machine LearningMultiagent SystemsRobotics
AI summaryⓘ
The authors studied how to control large groups of robots or agents (swarms) when the commands can only be sent occasionally, not continuously. Instead of trying to control their speed at every moment, they focus on controlling over fixed time intervals. They developed a learning method to find the best control strategy that uses minimal energy during each interval and showed how to train it efficiently. Their approach respects how real control systems work and makes steering large swarms easier with only a few updates.
swarm controlsampled-data systemslinear time-invariant dynamicsminimum-energy controlfinite-horizon controlcontrol-space learningbridge trajectoriesstop-gradient trainingscalable swarm steering
Authors
Anqi Dong, Yongxin Chen, Karl H. Johansson, Johan Karlsson
Abstract
Steering large-scale swarms in only a few control updates is challenging because real systems operate in sampled-data form: control inputs are updated intermittently and applied over finite intervals. In this regime, the natural object is not an instantaneous velocity field, but a finite-window control quantity that captures the system response over each sampling interval. Inspired by MeanFlow, we introduce a control-space learning framework for swarm steering under linear time-invariant dynamics. The learned object is the coefficient that parameterizes the finite-horizon minimum-energy control over each interval. We show that this coefficient admits both an integral representation and a local differential identity along bridge trajectories, which leads to a simple stop-gradient training objective. At implementation time, the learned coefficient is used directly in sampled-data updates, so the prescribed dynamics and actuation map are respected by construction. The resulting framework provides a scalable approach to few-step swarm steering that is consistent with the sampled-data structure of real control systems.