JZJun Zhang

Jun Zhang

*This is an algorithmically generated author record based on publicly available information.

Jun Zhang has a total of 8 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Top Cited Papers

    OverlapTransformer: An Efficient and Yaw-Angle-Invariant Transformer Network for LiDAR-Based Place Recognition

    by Junyi Ma, Jun Zhang, Jintao Xu, Rui Ai, Weihao Gu, Xieyuanli Chen

    Place recognition is an important capability for autonomously navigating vehicles operating in complex environments and under changing conditions. It is a key component for tasks such as loop closing in SLAM or global localization. In this letter, we address the problem of place recognition based on 3D LiDAR scans recorded by an autonomous vehicle. We propose a novel lightweight neural network exploiting the range image representation of LiDAR sensors to achieve fast execution with less than 2 ms per frame. We design a yaw-angle-invariant architecture exploiting a transformer network, which boosts the place recognition performance of our method. We evaluate our approach on the KITTI and Ford Campus datasets. The experimental results show that our method can effectively detect loop closures compared to the state-of-the-art methods and generalizes well across different environments. To evaluate long-term place recognition performance, we provide a novel dataset containing LiDAR sequences recorded by a mobile robot in repetitive places at different times.


  • Automating Shift-Scheduling Calibration by Using Bionic Optimization and Personalized Driver Models

    by Li Xu, Jun Zhang, Bin Shi, Wuqiang Meng

    Shift-scheduling calibration is important to the automobile industry, but it is repetitive and time-consuming; it is thus desirable to have a robot driver to automate this process. In this paper, we propose automating the calibration of shift-scheduling by using bionic optimization, i.e., particle swarm optimization (PSO), to guide the searching process and to integrate the driving styles into the calibration by equipping the robot drivers with personalized driver models. The personalized driver model is established by imitating the human driving behavior and is employed as a robot driver to conduct the driving cycle test, i.e., FTP-72 or US06, for candidate shifting schedules. The shifting performance is evaluated online via the computed performance index and/or AVL-Driver, regarding both driveability and fuel economy. Guided by PSO, candidate schedules are generated, tried, and evaluated until an optimal or near-optimal solution is obtained through iterations. Numerical experiments are presented to verify the feasibility and effectiveness of the proposed scheme. The shifting performance is improved by about 2% in computed performance index when compared with the base map. It is also suggested that personalized calibration is preferred if economically feasible.