Search papers, labs, and topics across Lattice.
AnchorVLA introduces an anchored diffusion policy for mobile manipulation that leverages a lightweight VLA backbone and a truncated diffusion schedule to efficiently generate multimodal actions. To address drift from action chunking, they incorporate a test-time self-correction mechanism with a residual correction module for per-step adjustments. Experiments across various mobile manipulation tasks demonstrate improved success and stability under disturbances and distribution shifts, while maintaining low-latency inference.
By cleverly anchoring diffusion sampling near plausible solutions and adding a lightweight residual correction, AnchorVLA achieves robust mobile manipulation with significantly reduced inference costs.
A central challenge in mobile manipulation is preserving multiple plausible action models while remaining reactive during execution. A bottle in a cluttered scene can often be approached and grasped in multiple valid ways. Robust behavior depends on preserving this action diversity while remaining reactive as the scene evolves. Diffusion policies are appealing because they model multimodal action distributions rather than collapsing to one solution. But in practice, full iterative denoising is costly at control time. Action chunking helps amortize inference, yet it also creates partially open-loop behavior, allowing small mismatches to accumulate into drift. We present AnchorVLA, a diffusion-based VLA policy for mobile manipulation built on the core insight that when sampling begins near a plausible solution manifold, extensive denoising is unnecessary to recover multimodal, valid actions. AnchorVLA combines a lightweight VLA adaptation backbone with an anchored diffusion action head, which denoises locally around anchor trajectories using a truncated diffusion schedule. This retains multimodal action generation while reducing inference cost for closed-loop control. Crucially, to mitigate chunking-induced drift, we introduce a test-time self-correction mechanism via a lightweight residual correction module that makes high-frequency, per-step adjustments during rollout. Across diverse mobile manipulation tasks, AnchorVLA improves success and stability under disturbances and distribution shifts while maintaining low-latency inference. The source code is made available at https://github.com/jason-lim26/AnchorVLA.
