CrossTracer: Cross-Embodiment Navigation via VLA Model Reasoning and Trace Residuals Adapting

2026-08-06

AI Summary

This work addresses the challenge that existing vision-language-action (VLA) models often neglect robot-specific motion constraints in cross-morphology navigation, yielding semantically plausible but physically infeasible trajectories. To bridge this gap, we propose CrossTracer, a hierarchical framework that introduces normalized image-plane waypoints as a unified interface between semantic reasoning and physical execution. CrossTracer integrates a pretrained VLA model, a vision-language trajectory proposer (VL-Tracer), and a morphology-aware adapter (CE-Adapter), enabling cross-morphology navigation in a shared pixel space through semantic trajectory proposal followed by residual refinement based on visual traversability and robot identity. Morphology-aware training trajectories are automatically generated via the annotation-free CE-RRT* algorithm. On the NaviTrace benchmark, CrossTracer achieves a score of 45.68, outperforming Gemini-2.5-Pro by 10.01 points (+28.1%), and demonstrates significantly improved success rates and efficiency in real-world deployments on both wheeled and legged robots.

Abstract

Vision-language-action (VLA) models provide strong semantic priors for robot navigation, but they often ignore embodiment-specific mobility constraints. A path that is semantically plausible for one robot may be physically infeasible for another. We propose CrossTracer, a hierarchical framework for cross-embodiment navigation through adaptive trace residuals. CrossTracer represents navigation plans as normalized image-plane waypoints, forming a unified pixel-space interface between semantic reasoning and physical grounding. First, Vision-Language Trace Proposer (VL-Tracer) adapts a pretrained VLA model to predict an initial navigation trace from egocentric observations and flexible goal specifications. Second, CE-Adapter refines this trace by predicting embodiment-conditioned residual corrections from visual traversability cues, robot identity, and the initial trace. To train the refinement module without costly manual annotation, Cross-Embodiment RRT* (CE-RRT*) converts panoptic segmentation into robot-conditioned traversability cost maps and generates cost-minimizing pixel-space traces. We evaluate CrossTracer on the NaviTrace benchmark, which tests whether a model can generate embodiment-consistent navigation traces from egocentric observations, language instructions, and robot embodiment types. CrossTracer achieves a total score of 45.68, outperforming the strongest evaluated general-purpose baseline, Gemini-2.5-Pro, by 10.01 points, corresponding to a 28.1% relative improvement. Real-world deployment on wheeled and legged robots further shows improved navigation success and execution efficiency.

Problem

• cross-embodiment navigation
• vision-language-action models
• mobility constraints
• navigation trace
• physical feasibility

Innovation

• cross-embodiment navigation
• vision-language-action models
• trace residuals
• pixel-space planning
• CE-RRT*