Dual-BEV Nav: Dual-layer BEV-based Heuristic Path Planning for Robotic Navigation in Unstructured Outdoor Environments

TL;DR

Problem: robust long-range navigation in unstructured outdoor environments with poor maps. Approach: Dual-BEV Nav uses LBPM for local BEV-based traversability and GBPM for global overhead-map hints, integrating them through path scoring. Contributions: introduction of Local BEV Perception Encoder, Task-driven Goal Decoder, depth-guided BEV features, trajectory-driven overhead-map training with Focal Loss and $L_{VIB}$ objective, and a unified cost-based selection mechanism. Findings: the method improves temporal-distance predictions and achieves 65 m navigation in real-world tests, demonstrating robustness in occluded or shifted maps. Significance: enables more reliable long-distance robotic navigation in challenging outdoor settings without relying on precise priors.

Abstract

Path planning with strong environmental adaptability plays a crucial role in robotic navigation in unstructured outdoor environments, especially in the case of low-quality location and map information. The path planning ability of a robot depends on the identification of the traversability of global and local ground areas. In real-world scenarios, the complexity of outdoor open environments makes it difficult for robots to identify the traversability of ground areas that lack a clearly defined structure. Moreover, most existing methods have rarely analyzed the integration of local and global traversability identifications in unstructured outdoor scenarios. To address this problem, we propose a novel method, Dual-BEV Nav, first introducing Bird's Eye View (BEV) representations into local planning to generate high-quality traversable paths. Then, these paths are projected onto the global traversability map generated by the global BEV planning model to obtain the optimal waypoints. By integrating the traversability from both local and global BEV, we establish a dual-layer BEV heuristic planning paradigm, enabling long-distance navigation in unstructured outdoor environments. We test our approach through both public dataset evaluations and real-world robot deployments, yielding promising results. Compared to baselines, the Dual-BEV Nav improved temporal distance prediction accuracy by up to $18.7\%$. In the real-world deployment, under conditions significantly different from the training set and with notable occlusions in the global BEV, the Dual-BEV Nav successfully achieved a 65-meter-long outdoor navigation. Further analysis demonstrates that the local BEV representation significantly enhances the rationality of the planning, while the global BEV probability map ensures the robustness of the overall planning.

Figures (6)

• Figure 1: Dual-layer BEV-based heuristic path planning for robotic navigation in unstructured outdoor environments: Dual-BEV Nav integrates traversability information from both local and global BEVs, providing the capability for target exploration and sequential target navigation. The Local BEV planning model identifies the traversability of the local scene. Based on overhead maps, the Global BEV planning model models a probability map of hints. The optimal path is obtained by mapping the local traversable waypoints onto the probability map.
• Figure 2: Task-driven goal decoder generates latent features based on task requirements, and by decoding the traversability feature and the latent feature, outputs the control information of the robot moving toward the sub-goal.
• Figure 3: Two BEV feature extraction methods. (a): Original overhead map. (b): Precise segmentation of traversability. (c): Probability map of the traversability hints, generated from GBPM.
• Figure 4: Training and inference architecture of GBPM. The robot's trajectory is used as the ground truth to constrain during training. When making predictions, the probability map is directly used as hints.
• Figure 5: Waypoints prediction visualization. (a): Waypoints prediction, the red line represents the available paths generated by the LBPM, while the green line indicates the optimal waypoints selected based on the GBPM. (b): Goal position observation.