Topologically Persistent Features-based Object Recognition in Cluttered Indoor Environments

TL;DR

This paper introduces a novel topological descriptor for 3D object recognition in cluttered indoor environments, effectively handling occlusions by capturing detailed shape features through persistent homology.

Contribution

It presents a slicing-based topological descriptor that improves recognition accuracy of occluded objects, outperforming existing deep learning methods.

Findings

Outperforms DGCNN and SimpleView in cluttered scene recognition

Effective in recognizing occluded objects in unseen environments

Uses persistent homology to capture detailed 3D shape features

Abstract

Recognition of occluded objects in unseen indoor environments is a challenging problem for mobile robots. This work proposes a new slicing-based topological descriptor that captures the 3D shape of object point clouds to address this challenge. It yields similarities between the descriptors of the occluded and the corresponding unoccluded objects, enabling object unity-based recognition using a library of trained models. The descriptor is obtained by partitioning an object's point cloud into multiple 2D slices and constructing filtrations (nested sequences of simplicial complexes) on the slices to mimic further slicing of the slices, thereby capturing detailed shapes through persistent homology-generated features. We use nine different sequences of cluttered scenes from a benchmark dataset for performance evaluation. Our method outperforms two state-of-the-art deep learning-based point cloud classification methods, namely, DGCNN and SimpleView.