GNN-PMB: A Simple but Effective Online 3D Multi-Object Tracker without Bells and Whistles

Multi-object tracking (MOT) is among crucial applications in modern advanced driver assistance systems (ADAS) and autonomous driving (AD) systems. The global nearest neighbor (GNN) filter, as the earliest random vector-based Bayesian tracking framework, has been adopted in most of state-of-the-arts trackers and widely accepted in the automotive industry. With the development of random finite set (RFS) theory, which facilitates a mathematically rigorous treatment of the MOT problem, different variants of RFS-based Bayesian filters have been developed. However, their usefulness in the real traffic for ADAS and AD application is still open to doubt. In this paper, it is first demonstrated that the latest RFS-based Bayesian tracking framework could be superior to typical random vector-based Bayesian tracking framework like GNN, via a systematic comparative study of both traditional random vector-based Bayesian filters with rule-based heuristic track maintenance and RFS-based Bayesian filters on the nuScenes validation dataset. Then, an RFS-based tracker, namely Poisson multi-Bernoulli filter using the global nearest neighbor (GNN-PMB), is proposed to LiDAR-based MOT tasks. This GNN-PMB tracker is simple to use but can achieve competitive results on the nuScenes dataset. Specifically, the proposed GNN-PMB tracker outperforms most of the state-of-the-art LiDAR-only trackers and LiDAR and camera fusion-based trackers, ranking the 3rd among all LiDAR-only trackers on nuScenes 3D tracking challenge leader board1 at the time of submission. Our code is available at here.