Bohuan Xue

Bohuan Xue

Auto-generated profile, based on publicly available information

Bohuan Xue

School of Data Science and Engineering, Xingzhi College, South China Normal University, Shanwei, China

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
18 March 2021 in IEEE Robotics and Automation Letters
Verified
Journal Article
Closed access
RFRui Fan
Rui Fan
HWHengli Wang
Hengli Wang
BXBohuan Xue
Bohuan Xue
HHHuaiyang Huang
Huaiyang Huang
YWYuan Wang
Yuan Wang
IPIoannis Pitas
Ioannis Pitas
This paper proposes three-filters-to-normal (3F2N), an accurate and ultrafast surface normal estimator (SNE), which is designed for structured range sensor data, e.g., depth/disparity images. 3F2N SNE computes surface normals by simply performing three filtering operations (two image gradient filters in horizontal and vertical directions, respectively, and a mean/median filter) on an inverse depth image or a disparity image (due to that disparity is in inverse proportion to depth). Despite the simplicity of 3F2N SNE, no similar method already exists in the literature. To evaluate the performance of our proposed SNE, we created three large-scale synthetic datasets (easy, medium and hard) using 24 3D mesh models, each of which is used to generate 1800--2500 pairs of depth images (resolution: 480x640 pixels) and the corresponding surface normal ground truth from different views. 3F2N SNE demonstrates the state-of-the-art performance, outperforming all other existing geometry-based SNEs, where the average angular errors with respect to the easy, medium and hard datasets are 1.6 degrees, 5.6 degrees and 15.3 degrees, respectively. Furthermore, our C++ and CUDA implementations achieve a processing speed of over 260 Hz and 21 kHz, respectively. We have also contributed our source code to the OpenCV repository. Our datasets and source code are publicly available at: sites.google.com/view/3f2n.

1 December 2019
Verified
Conference Paper
Open access
YZYilong Zhu
Yilong Zhu
BXBohuan Xue
Bohuan Xue
LZLinwei Zheng
Linwei Zheng
HHHuaiyang Huang
Huaiyang Huang
MLMing Liu
Ming Liu
RFRui Fan
Rui Fan
Localization, or position fixing, is an important problem in robotics research. In this paper, we propose a novel approach for long-term localization in a changing environment using 3D LiDAR. We first create the map of a real environment using GPS and LiDAR. Then, we divide the map into several small parts as the targets for cloud registration, which can not only improve the robustness but also reduce the registration time. We proposed a localization method called PointLocalization. PointLocalization allows us to fuse different kinds of odometers, which can optimize the accuracy and frequency of localization results. We evaluate our algorithm on an unmanned ground vehicle (UGV) using LiDAR and a wheel encoder, and obtain the localization results at more than 20 Hz after fusion. The algorithm can also localize the UGV in a 180-degree field of view (FOV). Using an outdated map captured six months ago, this algorithm shows great robustness, and the test results show that it can achieve an accuracy of 10 cm. PointLocalization has been tested for a period of more than six months in a crowded factory and has operated successfully over a distance of more than 2000 km.

16 May 2024 in IEEE/ASME Transactions on Mechatronics
Verified
Journal Article
Closed access
BYBowen Yang
Bowen Yang
JCJie Cheng
Jie Cheng
BXBohuan Xue
Bohuan Xue
JJJianhao Jiao
Jianhao Jiao
MLMing Liu
Ming Liu
Navigation in complex 3-D scenarios requires appropriate environment representation for efficient scene understanding and trajectory generation. We propose a highly efficient and extensible global navigation framework based on a tomographic understanding of the environment to navigate ground robots in multilayer structures. Our approach generates tomogram slices using the point cloud map to encode the geometric structure as ground and ceiling elevations. Then, it evaluates the scene traversability considering the robot's motion capabilities. Both the tomogram construction and the scene evaluation are accelerated through parallel computation. Our approach further alleviates the trajectory generation complexity compared with planning in 3-D spaces directly. It generates 3-D trajectories by searching through multiple tomogram slices and separately adjusts the robot height to avoid overhangs. We evaluate our framework in various simulation scenarios and further test it in the real world on a quadrupedal robot. Our approach reduces the scene evaluation time by three orders of magnitude and improves the path planning speed by three times compared with existing approaches, demonstrating highly efficient global navigation in various complex 3-D environments.

29 May 2023
Verified
Conference Paper
Open access
YFYi Feng
Yi Feng
BXBohuan Xue
Bohuan Xue
MLMing Liu
Ming Liu
QCQijun Chen
Qijun Chen
RFRui Fan
Rui Fan
Surface normal holds significant importance in visual environmental perception, serving as a source of rich geometric information. However, the state-of-the-art (SoTA) surface normal estimators (SNEs) generally suffer from an unsatisfactory trade-off between efficiency and accuracy. To resolve this dilemma, this paper first presents a superfast depth-to-normal translator (D2NT), which can directly translate depth images into surface normal maps without calculating 3D coordinates. We then propose a discontinuity-aware gradient (DAG) filter, which adaptively generates gradient convolution kernels to improve depth gradient estimation. Finally, we propose a surface normal refinement module that can easily be integrated into any depth-to-normal SNEs, substantially improving the surface normal estimation accuracy. Our proposed algorithm demonstrates the best accuracy among all other existing real-time SNEs and achieves the SoTA trade-off between efficiency and accuracy.

1 March 2024 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
BXBohuan Xue
Bohuan Xue
XYXiaoyang Yan
Xiaoyang Yan
Jin WuJin Wu
Jin Wu
JCJintao Cheng
Jintao Cheng
JJJianhao Jiao
Jianhao Jiao
Chengxi ZhangChengxi Zhang
Chengxi Zhang
Localization, an indispensable component in robotics and automation, encounters difficulties arising from appearance variations, resulting in inaccurate data associations. Semantic-based positioning mitigates these challenges by filtering out invalid data, such as moving vehicles and worn road markings. Building on this insight, we introduce a robust semantic visual localization system, which has been successfully deployed in real-world settings. The system employs neural networks to extract road markers and associate data with a semantic map. To enhance system reliability, we employ several data filters. These filters remove images that are prone to misrecognition or poorly processed by neural networks. We propose two techniques for vehicle state estimation. The first, utilizing the inverse perspective mapping (IPM) matrix, directly determines the vehicle’s central pose. The second technique derives the camera pose using the Perspective-N-Points (PnP) method and leverages external parameters to infer the vehicle’s central state. The wheel encoder, with its robust anti-noise capability, offers odometry in the absence of semantic information, enhancing the system’s resilience. The crux of our approach lies in distinct semantic strategies: utilizing lane lines for orientation and road markers exclusively for translation estimation. We also detail an automatic construction method for the semantic map, enhancing the system’s practicality. Experimental results indicate that the IPM method outperforms the PnP approach, leading to notably improved positioning accuracy. Additionally, the error distribution of the IPM method more closely aligns with a normal distribution compared to that of the PnP approach.

4 January 2023 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
ZZZuhao Zou
Zuhao Zou
LLLiang Li
Liang Li
XHXiangcheng Hu
Xiangcheng Hu
YZYilong Zhu
Yilong Zhu
BXBohuan Xue
Bohuan Xue
MLMing Liu
Ming Liu
Low-cost Inertial Measurement Units (IMUs) are ubiquitously used in the attitude estimation of cell phones and robots. Accurate and robust IMU calibration is required to ensure attitude estimation accuracy. This paper proposes an accurate and robust equipment-free IMU calibration method. We do not assume gyroscope biases are invariant during calibration compared with conventional methods because low-cost IMUs like MPU6000 have large gyroscope biases instability. We propose introducing biases removal and outlier-aware optimization to alleviate the impact of variant gyroscope biases. Furthermore, we introduce a multi-resolution analysis based static detector to detect subtle IMU motion in real data collection. Our detector can detect 84% subtle motions (1-degree rotations) present in the simulated calibration data while the conventional variance thresholding detector can only detect 31% of them. In addition, we derive a proper data collection method to guide the user to collect data effectively. We benchmark our method with another existing equipment-free method with synthetic and real datasets. The results in synthetic datasets show that our method is 25% more accurate and robust than the existing method. The results in real datasets vouch that our method achieves an estimation of IMU intrinsic parameters comparable to the ground truth. Furthermore, the roll and pitch estimation of MPU6000 using our calibration method are close to (< 0.15° in ±30°) that of an expensive factory-calibrated IMU in real testing.