Chengmin Zhou

Dr. Chengmin Zhou


Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

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

Representative Publications

View All Publications
25 November 2021 in Journal of Intelligent Manufacturing
Verified
Journal Article
Open access
PDF file
CZChengmin Zhou
Chengmin Zhou
BHBingding Huang
Bingding Huang
Pasi FräntiPasi Fränti
Pasi Fränti
Principles of typical motion planning algorithms are investigated and analyzed in this paper. These algorithms include traditional planning algorithms, classical machine learning algorithms, optimal value reinforcement learning, and policy gradient reinforcement learning. Traditional planning algorithms investigated includegraph search algorithms,sampling-based algorithms,interpolating curve algorithms, andreaction-based algorithms. Classical machine learning algorithms includemulticlass support vector machine,long short-term memory,Monte-Carlo tree searchandconvolutional neural network. Optimal value reinforcement learning algorithms includeQ learning,deep Q-learning network,double deep Q-learning network,dueling deep Q-learning network. Policy gradient algorithms includepolicy gradient method,actor-critic algorithm,asynchronous advantage actor-critic,advantage actor-critic,deterministic policy gradient,deep deterministic policy gradient,trust region policy optimizationandproximal policy optimization. New general criteria are also introduced to evaluate the performance and application of motion planning algorithms by analytical comparisons. The convergence speed and stability of optimal value and policy gradient algorithms are specially analyzed. Future directions are presented analytically according to principles and analytical comparisons of motion planning algorithms. This paper provides researchers with a clear and comprehensive understanding about advantages, disadvantages, relationships, and future of motion planning algorithms in robots, and paves ways for better motion planning algorithms in academia, engineering, and manufacturing.

2 March 2023 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Open access
CZChengmin Zhou
Chengmin Zhou
BHBingding Huang
Bingding Huang
Pasi FräntiPasi Fränti
Pasi Fränti
Indoor motion planning challenges researchers because of the high density and unpredictability of moving obstacles. Classical algorithms work well in the case of static obstacles but suffer from collisions in the case of dense and dynamic obstacles. Recent reinforcement learning (RL) algorithms provide safe solutions for multiagent robotic motion planning systems. However, these algorithms face challenges in convergence: slow convergence speed and suboptimal converged result. Inspired by RL and representation learning, we introduced the ALN-DSAC: a hybrid motion planning algorithm where attention-based long short-term memory (LSTM) and novel data replay combine with discrete soft actor–critic (SAC). First, we implemented a discrete SAC algorithm, which is the SAC in the setting of discrete action space. Second, we optimized existing distance-based LSTM encoding by attention-based encoding to improve the data quality. Third, we introduced a novel data replay method by combining the online learning and offline learning to improve the efficacy of data replay. The convergence of our ALN-DSAC outperforms that of the trainable state of the arts. Evaluations demonstrate that our algorithm achieves nearly 100% success with less time to reach the goal in motion planning tasks when compared to the state of the arts. The test code is available at https://github.com/CHUENGMINCHOU/ALN-DSAC.

13 June 2018 in Journal of Intelligent & Fuzzy Systems: Applications in Engineering and Technology
Verified
Journal Article
Closed access
PDF file
CZChengmin Zhou
Chengmin Zhou
FLFei Li
Fei Li
WCWen Cao
Wen Cao
CWCao Wang
Cao Wang
YWYihuai Wu
Yihuai Wu
Contrasted with common obstacle avoidance mode based on single sensor or solo algorithm, this article put forward an intelligent pattern based on Combination from CNN-based Deep Learning Method and liDAR-based Image Processing approach. As for Deep Learning method, a 10-layer Convolutional Neural Network (CNN) is designed which comes to a high recognition accuracy of 97 percent in Tensorflow and success rate of obstacle avoidance is over 90 percent. With regard to liDAR-based Image Processing approach, decision is made by a special method of counting the number of Point Cloud Data (PCD) which is generated by 2D liDAR and a success rate over 90 percent is achieved as well. When two kinds of methods work together, a robust success rate of 100 percent is realized. Meanwhile, Inertial Measurement Unit (IMU) and Xbox360 are taken into consideration for Pose Estimation and Data Collection. Finally, all functions are integrated in Robot Operation System (ROS) on platform of nVidia Jetson TX1.

24 March 2020 in Robotics
Verified
Journal Article
Open access
PDF file
CLChunmei Liu
Chunmei Liu
CZChengmin Zhou
Chengmin Zhou
CWcao wen
cao wen
FLFei Li
Fei Li
Pengfei JiaPengfei Jia
Pengfei Jia
Pervasive deployment of autonomous vehicle all over the world is an undisputed trend in the future. Autonomous vehicle will inevitably play an essential role in decreasing traffic jams, reducing threats from driving while intoxicated (DWI), and assisting the handicapped to get around. At the same time, the new energy vehicles (NEV) especially the electromobile is gradually adopted by several governments like Germany, USA and China as compulsive transportation tools from the standpoint of environmental friendliness. Taking these two crucial trends into consideration, this article proposes a scheme of autonomous robotic car based on robot operation system (ROS) in electromobile-like car which can be easily transplanted to commercial electromobile. In this article, the design and implementation of robotic car are demonstrated in detail which involves overall architecture of functional modules, hardware design, obstacle avoidance, localization and mapping, land detection and tracking, velocity control and indoor navigation. All software modules and hardware are integrated in NVIDIA Jetson TX1 and TRAXXAS car.

3 March 2018 in Multimedia Tools and Applications
Verified
Journal Article
Closed access
PDF file
CZChengmin Zhou
Chengmin Zhou
FLFei Li
Fei Li
WCWen Cao
Wen Cao

15 August 2025 in Robotics and Autonomous Systems
Verified
Journal Article
Open access
CZChengmin Zhou
Chengmin Zhou
XLXin Lu
Xin Lu
JDJiapeng Dai
Jiapeng Dai
XLXiaoxu Liu
Xiaoxu Liu
BHBingding Huang
Bingding Huang
Pasi FräntiPasi Fränti
Pasi Fränti