Inverse Safety Filtering: Inferring Constraints from Safety Filters for Decentralized Coordination

TL;DR

This paper presents an online method for inferring safety constraints from observed actions in multi-agent systems, enabling decentralized safety-aware coordination without explicit communication.

Contribution

It introduces a novel constraint inference approach from safety-filtered actions, with convergence guarantees and practical validation on robots.

Findings

The method accurately infers constraints in simulations.

It ensures safety in decentralized planning with large activation distances.

Validated on quadruped robot experiments.

Abstract

Safe multi-agent coordination in uncertain environments can benefit from learning constraints from other agents. Implicitly communicating safety constraints through actions is a promising approach, allowing agents to coordinate and maintain safety without expensive communication channels. This paper introduces an online method to infer constraints from observing the safety-filtered actions of other agents. We approach the problem by using safety filters to ensure forward safety and exploit their structure to work backwards and infer constraints. We provide sufficient conditions under which we can infer these constraints and prove that our inference method converges. This constraint inference procedure is coupled with a decentralized planning method that ensures safety when the constraint activation distance is sufficiently large. We then empirically validate our method with Monte Carlo simulations and hardware experiments with quadruped robots.