AI & ML
Distilling Safety Filters for Filter-free Multi-agent RL
arXiv cs.RO0 views
arXiv:2611.19190v1 Announce Type: new Abstract: Reinforcement learning (RL) can teach robot teams to navigate shared spaces, but trial and error causes collisions that real robots cannot afford. Control barrier function (CBF) safety filters prevent these collisions by correcting unsafe actions, yet in a team the filter becomes a centralized optimization whose number of constraints grows quadratically with the team size. We propose an approach that builds the filter's safety into the RL policy and examine how such a system behaves during filter-free deployment. However, the filter that keeps training safe also hides what the policy must learn: it masks the cost of unsafe actions, which actor-critic methods then overvalue, and it keeps near-collision states out of the training data. Our approach therefore learns only from executed safe actions with implicit Q-learning and distills the filter into the policy at synthetic near-collision states. We prove that filter distillation loss bounds the policy's constraint violation and hence how far agents can cross their safety margin, so errors never accumulate. When distillation states cover deployed states, the loss also bounds the collision probability at every step. Without a filter and acting only on local observations, our policies have no collisions in 1,800 test episodes with multiple agents. A policy trained with three agents transfers zero-shot to teams of twenty at constant per-agent computation and stays collision-free with up to eight agents. We test our policies on Crazyflie quadrotors with up to five flying at the same time.
Source:https://arxiv.org/abs/2611.19190PreviousImpedanceMimic: Learned Impedance as Action Space for Force-Accurate Motion ImitationNextEgo2Sim2Humanoid: Agentic Real-to-Sim for Learning Loco-Manipulation VLAs without Robot Demonstrations
Related posts
Discontinuity-Aware Solvers for Vapor-Compression Cycles
0arXiv cs.NA
Memorization capacity of deep ReLU neural networks characterized by width and depth
0arXiv cs.NA
Time-Delayed Transformers for Data-Driven Modeling of Low-Dimensional Dynamics
0arXiv cs.NA
First-order methods for stochastic and finite-sum convex optimization with deterministic constraints
0arXiv cs.NA
Smooth Surfaces via Nets of Geodesics
0arXiv cs.NA
A Large Gap between Galerkin Errors with and without Mesh-Uniform Regularity for the KdV Equation with Boundary Dissipation
0arXiv cs.NA