Collaborative Multi-Object Tracking with Conformal Uncertainty Propagation

• DOI: 10.48550/arxiv.2303.14346
• Publication Date: 2023-01-01
• Authors: Sanbao Su; Songyang Han; Yiming Li; Zhili Zhang; Feng Chen; Caiwen Ding; Fei Miao
• Keywords: Leverage (statistics); Robustness; Uncertainty reduction theory; Viewpoints; Propagation of uncertainty

Object detection and multiple object tracking (MOT) are essential components of self-driving systems. Accurate uncertainty quantification both critical for onboard modules, such as perception, prediction, planning, to improve the safety robustness autonomous vehicles. Collaborative (COD) has been proposed accuracy reduce by leveraging viewpoints agents. However, little attention paid how leverage from COD enhance MOT performance. In this paper, first attempt address challenge, we design an propagation framework called MOT-CUP. Our quantifies through direct modeling conformal propagates information into motion prediction association steps. MOT-CUP is designed work with different collaborative detectors baseline algorithms. We evaluate on V2X-Sim, a comprehensive perception dataset, demonstrate 2% improvement in 2.67X reduction compared baselines, e.g. SORT ByteTrack. scenarios characterized high occlusion levels, our demonstrates noteworthy $4.01\%$ accuracy. importance MOT, provides based propagation. code public https://coperception.github.io/MOT-CUP/.