MPCoT: Reward-Guided Multi-Path Latent Reasoning for Test-Time Scalable Vision-Language-Action
This work addresses the brittleness of VLA policies in long-horizon and uncertain control tasks by enabling test-time scalable reasoning without token overhead, which is important for robotics and embodied AI.
MPCoT introduces a multi-path latent reasoning framework for Vision-Language-Action policies that improves long-horizon control performance by initializing multiple hypotheses, refining them through weight-tied steps, and softly aggregating them without generating explicit reasoning tokens. On LIBERO and CALVIN benchmarks, it achieves better long-horizon results compared to baselines.
Vision-Language-Action (VLA) policies remain brittle in long-horizon and high-uncertainty control, where one-pass action decoding provides limited inference-time deliberation. Explicit chain-of-thought can increase reasoning depth, but introduces token latency and an indirect text-to-action interface. We propose MPCoT, a reward-guided multi-path latent reasoning framework that initializes $M$ hypotheses, refines them for K weight-tied steps, and softly aggregates them before action decoding. A training-only path-preference objective evaluates candidate action branches with expert-action consistency, world-model/VLM-based progress, and success feedback to align the latent path scorer with downstream execution quality. MPCoT preserves the original 8-step action interface, generates zero reasoning tokens, and exposes configurable inference controls (K,M). Under matched protocols on LIBERO and CALVIN, MPCoT improves long-horizon performance, with ablations confirming depth-width effects, confidence-weighted aggregation, and reward-guided path supervision.