Cross-Vehicle 3D Geometric Consistency for Self-Supervised Surround Depth Estimation on Articulated Vehicles

TL;DR

This paper introduces ArticuSurDepth, a self-supervised surround-view depth estimation framework tailored for articulated vehicles, leveraging geometric consistency and structural priors to improve 3D perception in autonomous driving.

Contribution

It presents a novel self-supervised method that incorporates cross-view and cross-vehicle geometric constraints, specifically designed for articulated vehicles, with validation on a new dataset and benchmarks.

Findings

Achieved state-of-the-art depth estimation performance on multiple datasets.

Enhanced structural coherence through multi-view spatial context and surface normal constraints.

Demonstrated effectiveness on a newly established articulated vehicle dataset.

Abstract

Surround depth estimation provides a cost-effective alternative to LiDAR for 3D perception in autonomous driving. While recent self-supervised methods explore multi-camera settings to improve scale awareness and scene coverage, they are primarily designed for passenger vehicles and rarely consider articulated vehicles or robotics platforms. The articulated structure introduces complex cross-segment geometry and motion coupling, making consistent depth reasoning across views more challenging. In this work, we propose \textbf{ArticuSurDepth}, a self-supervised framework for surround-view depth estimation on articulated vehicles that enhances depth learning through cross-view and cross-vehicle geometric consistency guided by structural priors from vision foundation model. Specifically, we introduce multi-view spatial context enrichment strategy and a cross-view surface normal constraint to improve structural coherence across spatial and temporal contexts. We further incorporate camera height regularization with ground plane-awareness to encourage metric depth estimation, together with cross-vehicle pose consistency that bridges motion estimation between articulated segments. To validate our proposed method, an articulated vehicle experiment platform was established with a dataset collected over it. Experiment results demonstrate state-of-the-art (SoTA) performance of depth estimation on our self-collected dataset as well as on DDAD, nuScenes, and KITTI benchmarks.