Search papers, labs, and topics across Lattice.
This paper introduces NavCMPO, a two-stage adaptive navigation framework that enhances mapless visual navigation by integrating few-step MeanFlow trajectory generation with critic-guided refinement and reinforcement learning fine-tuning. The approach addresses the inference latency and performance limitations of existing methods by employing an obstacle proximity prediction task during pre-training and refining trajectories using gradients from a critic trained with obstacle-point-cloud supervision. Results show that NavCMPO achieves a 74.7% success rate on the InternVLA-N1 benchmark, outperforming the NavDP baseline by 6.4 percentage points while significantly reducing inference latency from 85 ms to 60 ms.
NavCMPO not only boosts navigation success rates but also slashes inference latency, making real-time obstacle avoidance more feasible in complex environments.
End-to-end diffusion-based policies have demonstrated strong performance in mapless visual navigation, but their iterative denoising process introduces substantial inference latency, while behavior cloning limits performance to the quality of expert demonstrations. We present NavCMPO, a two-stage adaptive navigation framework that combines few-step MeanFlow trajectory generation, critic-guided refinement, and reinforcement learning fine-tuning. During pre-training, an obstacle proximity prediction task encourages the visual representation to capture obstacle-aware spatial information. To compensate for the degradation in obstacle avoidance caused by few-step generation, Critic-Guided Trajectory Refinement (CGTR) uses gradients from a critic trained with obstacle-point-cloud supervision to refine intermediate trajectories. During adaptation, the MeanFlow policy is fine-tuned using Proximal Policy Optimization with behavior-cloning regularization, while the critic is updated to accommodate embodiment-specific observation changes. Under a matched training budget on the InternVLA-N1 benchmark, NavCMPO achieves an average success rate of 74.7\%, exceeding the retrained NavDP baseline by 6.4 percentage points, while reducing inference latency from 85\,ms to 60\,ms. Experiments on a Unitree Go2 further demonstrate effective sim-to-real transfer.
