Robust Dynamic Object Detection in Cluttered Indoor Scenes via Learned Spatiotemporal Cues

TL;DR

This paper introduces a LiDAR-only dynamic object detection method that combines motion segmentation with learned spatiotemporal priors, significantly improving recall and F1 score in cluttered indoor scenes for autonomous navigation.

Contribution

The work presents a novel fusion framework that integrates occupancy-grid motion segmentation with a learned BEV dynamic prior, enhancing detection robustness in cluttered environments.

Findings

28.67% higher recall than state-of-the-art

18.50% higher F1 score

Maintains comparable precision and position error

Abstract

Reliable dynamic object detection in cluttered environments remains a critical challenge for autonomous navigation. Purely geometric LiDAR pipelines that rely on clustering and heuristic filtering can miss dynamic obstacles when they move in close proximity to static structure or are only partially observed. Vision-augmented approaches can provide additional semantic cues, but are often limited by closed-set detectors and camera field-of-view constraints, reducing robustness to novel obstacles and out-of-frustum events. In this work, we present a LiDAR-only framework that fuses temporal occupancy-grid-based motion segmentation with a learned bird's-eye-view (BEV) dynamic prior. A fusion module prioritizes 3D detections when available, while using the learned dynamic grid to recover detections that would otherwise be lost due to proximity-induced false negatives. Experiments with motion-capture ground truth show our method achieves 28.67% higher recall and 18.50% higher F1 score than the state-of-the-art in substantially cluttered environments while maintaining comparable precision and position error.