Planning Oriented 3D Scene Completion via Coupled TUDF Occupancy Representation Learning from Partial Observations

2026-09-28

AI Summary

This study addresses the challenge of incomplete observations caused by sensor occlusion in robot navigation, where existing completion methods struggle to directly support path planning. We propose a planning-oriented 3D scene completion framework that introduces a novel bidirectional coupled learning mechanism between Truncated Unsigned Distance Fields (TUDF) and voxel occupancy maps based on partial LiDAR data. Specifically, TUDF guides occupancy reconstruction while occupancy features reciprocally refine distance field estimation, thereby enhancing obstacle boundary reasoning. This joint representation enables seamless integration of geometric completion and trajectory planning. Experiments demonstrate that our approach significantly improves both reconstruction quality and planning performance in unseen environments, allowing direct application without requiring post-processing.

Abstract

Partial observability remains a fundamental challenge in robotic navigation, where limited sensor coverage and occlusions leave large portions of the environment unobserved. Existing scene completion methods primarily focus on improving incomplete mapping or reconstructing partially observed 3D structures, but rarely investigate how scene completion can be designed to benefit downstream tasks such as path planning. In this work, we propose a path-planning-oriented 3D scene completion framework that moves beyond pure occupancy modeling toward a coupled geometric formulation. Specifically, given partial LiDAR observations as input, the proposed framework jointly predicts completed Truncated Unsigned Distance Field (TUDF)-based continuous geometric representations and voxel-wise occupancy maps. This coupled representation allows the network to better reason about obstacle boundaries and free-space geometry. To fully exploit the synergy between the two representations, we introduce a bidirectionally coupled learning scheme, where TUDF features provide dense geometric guidance to improve occupancy reconstruction, while occupancy features in turn offer complementary structural constraints that refine distance-field estimation. Consequently, the proposed network directly predicts complete occupancy and TUDF representations, allowing seamless integration of TUDF into trajectory planning without post-processing. Extensive experiments on unseen environments demonstrate that the proposed method consistently improves both geometric reconstruction quality and downstream planning performance.

Problem

• 3D Scene Completion
• Robotic Navigation
• Path Planning
• Partial Observability

Innovation

• 3D Scene Completion
• Truncated Unsigned Distance Field (TUDF)
• Coupled Representation Learning
• Path Planning
• Partial Observability