The paper examines how transformer policies, which process agents as ordered token sequences, perform in multi‑agent robot learning where agent teams are unordered. It finds that low permutation error can mask action collapse, where all agents choose the same action, and proposes additional diagnostics such as action diversity and same‑action fraction. Experiments show that while a PPO‑ID baseline avoids collapse, it remains order‑sensitive, and that strong equivariance regularization can still cause homogeneous behavior; a weak penalty improves robustness and preserves diversity for three‑agent teams, but four‑agent teams need much smaller regularization weights.
arXiv:2607. 25132v2 Announce Type: replace Abstract: A central challenge in interpreting learned decision-making systems is to determine whether their internal representations contain concepts that help explain their behavior.
The paper introduces Distributed Dexterous Manipulation (DDM), a challenging control problem involving 64 soft delta robots arranged in an 8x8 grid. It presents a framework using spatially conditioned Multi-Agent Transformers (MATs) with adaptive layer norm, spatial contrastive embeddings, and a behavior cloning method fine‑tuned by Soft Actor Critic. Experiments demonstrate that MATs refine actions through stacked attention blocks, enabling long‑horizon planar manipulation in simulation and real‑world settings, while an action‑selection strategy reduces robot usage by about 65% and lowers wear‑and‑tear, achieving an average error of ~1.5 cm.
arXiv:2609.36588v1 Announce Type: cross Abstract: We study reinforcement learning (RL) methods for cooperative multi-agent Vision-Language-Action (VLA) models. This problem is challenging because VLA...
arXiv:2604. 13472v2 Announce Type: replace-cross Abstract: Cooperative multi-agent reinforcement learning (MARL) is widely used to address large joint observation and action spaces by decomposing a centralized control problem into multiple interacting agents.
Vision-Language-Action (VLA) models have attracted growing interest as a scalable approach to robotic manipulation. While these models are effective action predictors, deploying them as robotic agents exposes critical gaps: no mechanism for failure recovery, inconsistent execution over long horizons, and limited robustness to shifts in observations, tasks, or embodiments.
arXiv:2608. 11658v1 Announce Type: cross Abstract: Many reinforcement learning systems, from fleet management to traffic signal control, must serve an objective that changes dynamically after deployment, and retraining a policy for each new objective is prohibitively expensive.
arXiv:2508. 20072v4 Announce Type: replace-cross Abstract: Vision-Language-Action (VLA) models adapt large vision-language backbones to map images and instructions into robot actions.
arXiv:2607. 21670v1 Announce Type: cross Abstract: Action tokenization maps continuous robot action chunks to discrete tokens and has become an important interface for modern visuomotor policies.
arXiv:2607. 17914v1 Announce Type: cross Abstract: Robust multi-agent coordination relies heavily on inter-agent communication, which is frequently disrupted by physical and environmental constraints in real-world deployments.
The paper investigates a specific failure mode in vision‑language‑action models, termed instruction‑action binding, where models respond to language and vision separately but fail to combine them to select the correct action under counterfactual changes. Through behavioral analyses of fine‑tuned policies, the authors show that failed rollouts often preserve source behavior or switch to other demonstrated tasks, indicating that language is not ignored but mis‑bound. They propose Equivariant Counterfactual Training (ECT), which supplies counterfactual demonstrations and a paired loss to enforce correct action selection, achieving significant performance gains across simulated and real‑world benchmarks.
arXiv:2606. 30072v1 Announce Type: new Abstract: Cooperative tasks in Multi-Agent Reinforcement Learning (MARL) require agents to collectively maximize a shared return.
arXiv:2512. 05693v2 Announce Type: replace-cross Abstract: Generalist vision--language--action (VLA) policies are typically trained on heterogeneous mixtures of robot demonstrations spanning diverse embodiments, action spaces, and observation configurations.