SDT-6D: Fully Sparse Depth-Transformer for Staged End-to-End 6D Pose Estimation in Industrial Multi-View Bin Picking

TL;DR

This paper presents SDT-6D, a fully sparse depth-transformer framework for precise 6D pose estimation in cluttered industrial environments, leveraging multi-view depth fusion, scene-adaptive attention, and a novel voting strategy.

Contribution

It introduces a staged heatmap mechanism and a density-aware sparse transformer for high-resolution, sparse 3D data processing in 6D pose estimation.

Findings

Achieves competitive accuracy on IPD and MV-YCB datasets.

Effectively handles occlusions and clutter in industrial bin picking.

Operates efficiently with high-resolution sparse volumetric data.

Abstract

Accurately recovering 6D poses in densely packed industrial bin-picking environments remain a serious challenge, owing to occlusions, reflections, and textureless parts. We introduce a holistic depth-only 6D pose estimation approach that fuses multi-view depth maps into either a fine-grained 3D point cloud in its vanilla version, or a sparse Truncated Signed Distance Field (TSDF). At the core of our framework lies a staged heatmap mechanism that yields scene-adaptive attention priors across different resolutions, steering computation toward foreground regions, thus keeping memory requirements at high resolutions feasible. Along, we propose a density-aware sparse transformer block that dynamically attends to (self-) occlusions and the non-uniform distribution of 3D data. While sparse 3D approaches has proven effective for long-range perception, its potential in close-range robotic applications remains underexplored. Our framework operates fully sparse, enabling high-resolution volumetric representations to capture fine geometric details crucial for accurate pose estimation in clutter. Our method processes the entire scene integrally, predicting the 6D pose via a novel per-voxel voting strategy, allowing simultaneous pose predictions for an arbitrary number of target objects. We validate our method on the recently published IPD and MV-YCB multi-view datasets, demonstrating competitive performance in heavily cluttered industrial and household bin picking scenarios.