CoWAM: Coordination Contracts for Selective Policy Intervention with WAMs

2026-08-03

AI Summary

This work addresses the challenge of enabling selective intervention in dual-arm robotic tasks while preserving motion coordination and safety. The authors propose CoWAM, a coordinated contract-based intervention layer that formalizes coordination contracts—such as synchronicity, role compatibility, and collision convergence—as verifiable intervention conditions for the first time. By integrating a calibrated intervention gate with an abstention fallback mechanism, CoWAM selectively replaces nominal actions only under low-risk conditions. Leveraging World Action Models, typed acceptability checks, and decoupled decision-making over a shared candidate pool, CoWAM achieves a 16.7 percentage point improvement in coordinated effective selection rate and a 9.6 percentage point gain in closed-loop task success across eight simulated tasks, while maintaining a harmful intervention rate below 1%.

Abstract

World Action Models (WAMs) augment robot policies with action-conditioned predicted futures, but a plausible future alone does not justify changing the action that a bimanual policy would execute. We present CoWAM, a selective intervention layer that expresses synchronization, role compatibility, and collision convergence as coordination contracts. Each contract combines typed admissibility checks with event-conditioned verification and calibrated intervention gates. CoWAM preserves the nominal action unless an alternative satisfies every active obligation and provides a clear, low-risk improvement; when the nominal action is also inadmissible, it invokes a predefined abstention fallback. To separate selector quality from proposal quality, all methods operate on identical candidate pools and commit their decisions before shared oracle labeling. Across eight simulated bimanual tasks, CoWAM improves coordination-valid selection by 16.7 percentage points over the contract-only variant and raises closed-loop success by 9.6 percentage points over the strongest selective baseline, while keeping harmful interventions below 1%. Together, these results establish coordination contracts as an effective interface for conservative policy intervention with predicted world-action evidence across coordination-rich bimanual tasks.

Problem

• bimanual coordination
• policy intervention
• coordination contracts
• World Action Models
• selective intervention

Innovation

• coordination contracts
• selective intervention
• World Action Models
• bimanual coordination
• policy augmentation