VQActFlow: Vector-Quantized Action Mode Steering for Multi-Task Robot Manipulation
For multi-task robot manipulation, VQActFlow addresses the challenge of action mode selection from multimodal demonstration distributions, improving task success and feasibility.
VQActFlow tokenizes continuous robot actions into a discrete codebook and uses Variational Flow Matching to generate code sequences, enabling explicit mode preference steering via inference-time guidance. It achieves state-of-the-art performance on LIBERO simulation benchmarks and real-world humanoid and bimanual platforms, outperforming both continuous and discrete baselines.
Multi-task robot manipulation policies are challenging to learn from demonstration because traditionally a single network must select among qualitatively different action modes from a multimodal demonstration distribution, conditioned on language and visual context. A wrong mode selection means executing the wrong task or an action infeasible in the scene. Tokenizing continuous actions into a learned discrete codebook separates these modes at the representation level, offering structural advantages for multi-task learning. We propose VQActFlow, a multi-task manipulation policy that tokenizes action chunks and generates code sequences via Variational Flow Matching. VQActFlow maintains an explicit preference over action modes throughout generation. Inference-time guidance acts on this preference to steer mode commitment. We instantiate this with classifier-free guidance over language conditioning, which steers the policy toward the instructed action mode, and a learned codebook critic that supplies a complementary feasibility signal. We evaluate VQActFlow on three platforms: the LIBERO simulation benchmarks, a Unitree G1 humanoid performing whole-body pick-and-place, and an ALOHA-style bimanual platform performing contact-rich tasks. Across these benchmarks, VQActFlow outperforms both continuous and discrete baselines.