1: \begin{abstract}

2:   % (e.g., occlusion, motion blur, or light weakness),

3:   3D multiple object tracking (MOT) plays a crucial role in autonomous driving perception. 

4:   Recent end-to-end query-based trackers simultaneously detect and track objects, which have shown promising potential for the 3D MOT task.

5:   However, existing methods overlook the \textit{uncertainty} issue, which refers to the lack of precise confidence about the state and location of tracked objects. 

6:   % Uncertainty arises owing to various factors during motion observation by cameras, 

7:   Uncertainty arises owing to various factors during camera observation, especially occlusions and the small size of target objects, resulting in an inaccurate estimation of the object's position, label, and identity.

8:   To this end, we propose an \textbf{U}ncertainty-\textbf{A}ware 3D MOT framework, \name{}, which tackles the uncertainty problem from multiple aspects. 

9:   Specifically, we first introduce an Uncertainty-aware Probabilistic Decoder to capture the uncertainty in object prediction with probabilistic attention. 

10:   Secondly, we propose an Uncertainty-guided Query Denoising strategy 

11:   % to further enhance the training process.

12:   to enhance training robustness and convergence against uncertainty.

13:   We also utilize Uncertainty-reduced Query Initialization, which leverages predicted 2D object location and depth information to 

14:   reduce query uncertainty. 

15:   % reduce uncertainty in less biased initial queries.

16:   As a result, our \name{} achieves state-of-the-art performance on nuScenes benchmark, i.e., 66.3\% AMOTA on test split, surpassing the previous best end-to-end solution by a significant margin of 8.9\% AMOTA.

17: 

18:   \keywords{\name{} \and Uncertainty-Aware \and 3D Multi-Object Tracking  }

19: \end{abstract}

20: