Suboptimal event-triggered control over unreliable communication links with experimental validation

Abstract

We propose an event-triggered control policy for a discrete-time linear system with unreliable actuators' and sensors' links, captured by Bernoulli packet dropout models. The proposed policy is a threshold-based policy by which transmissions occur if a weighted norm of an error state vector exceeds a threshold. The threshold and the weights of the norm depend on the underlying characteristics of the packet dropout model. Such a policy is shown to guarantee a given performance, defined in terms of a quadratic cost, while reducing transmissions. The proposed event-triggered control policy is experimentally validated in the context of remotely steering an omni-directional ground robot along a predefined trajectory over an unreliable wireless network, while keeping transmissions to a minimum.