Dr. Guangda Chen received his Ph.D. degree in Computer Science from University of Science and Technology of China in 2021 (BA17011, advised by Prof. Xiaoping Chen in the USTC Robotics Laboratory). Before joining USTC Robotics Lab in 2015, he received a Bachelor of Administration from China Medical University in 2014. His research focuses on mobile robot navigation in dynamic and crowded environments, and interests include Reinforcement Learning, Computer Vision and Sensor Calibration.

Find him on:

University of Science and Technology of China, PhD
- School of Computer Science and Technology, 2015 — 2021

China Medical University, BAdmin
- Medical Informatics, 2011 — 2014
- Stomatology, 2010 — 2011

Panda Robot (2020 - now)
Tour guide robot at the Chengdu Giant Panda Museum.
Group Leader, (PhD thesis, Ch. 4 and Ch. 5)

• Pedestrian detection and tracking, [Demo]
• Long-term robust localization, [Demo]
• Navigation in dense crowds, [Brief Demo], [Full Demo]
• Tour guide service v1.0, [Demo]

DRL-based Navigation (2019 - now)
Group Leader , Open source code on GitHub

• DQN-based Obstacle Avoidance [ref.8, 9], [Demo]
• DRQN-based 3D Obstacle Avoidance [ref.7], [Demo]
• Multi-Robot Collision Avoidance [ref.5, 6], [Demo]
• Crowd Navigation [ref.3], [Demo], [Code statistics]
• Crowd Navigation with Interaction Capacity [ref.2]

Kejia Robot (2015 - 2019)
Team: WrightEagle@Home
[TDP ], [Poster ], [Report ], [Demo]

for Testing:
- Cleaning Robots Test
- Anhui Robot Technology Standard Innovation Base
for Calibration:
- General Batch-Calibration Framework
- RGB-D Cameras Calibration [ref.10]
for Training:
- Real Simulation Unified Platform

All publications: [Google Scholar] [ORCID] [Publons] [ResearchGate]

1. Robot Navigation in Complex and Dynamic Pedestrian Scenarios
   复杂动态行人场景下的机器人导航
   Guangda Chen. [PhD Thesis]
   University of Science and Technology of China. Hefei, China. June, 2021.
   [BibTeX], [PDF], [Slides]

   @phdthesis{cgd-phd,
   title = {复杂动态行人场景下的机器人导航},
   author = {陈广大},
   year = {2021},
   school = {中国科学技术大学},
   ADDRESS = "合肥"
   };

---

2. Learning to Socially Navigate in Pedestrian-rich Environments with Interaction Capacity
   Quecheng Qiu, Shunyi Yao , Jing Wang , Jun Ma , Guangda Chen and Jianmin Ji.
   ICRA 2022, Accepted
   [Abstract]
   

   Abstract: Existing navigation policies for autonomous robots tend to focus on collision avoidance while ignoring human-robot interactions in social life. For instance, robots can pass along the corridor safer and easier if pedestrians notice them. Sounds have been considered as an efficient way to attract the attention of pedestrians, which can alleviate the freezing robot problem. In this work, we present a new deep reinforcement learning (DRL) based social navigation approach for autonomous robots to move in pedestrian-rich environments with interaction capacity. Most existing DRL based methods intend to train a general policy that outputs both navigation actions, i.e., expected robot's linear and angular velocities, and interaction actions, i.e., the beep action, in the context of reinforcement learning. Different from these methods, we intend to train the policy via both supervised learning and reinforcement learning. In specific, we first train an interaction policy in the context of supervised learning, which provides a better understanding of the social situation, then we use this interaction policy to train the navigation policy via multiple reinforcement learning algorithms. We evaluate our approach in various simulation environments and compare it to other methods. The experimental results show that our approach outperforms others in terms of the success rate. We also deploy the trained policy on a real-world robot, which shows a nice performance in crowded environments

---

3. Crowd-Aware Robot Navigation for Pedestrians with Multiple Collision Avoidance Strategies via Map-based Deep Reinforcement Learning
   Shunyi Yao∗, Guangda Chen∗, Quecheng Qiu, Jun Ma, Xiaoping Chen and Jianmin Ji.
   IROS 2021
   [Abstract], [Demo], [PDF]
   

   Abstract: It is challenging for a mobile robot to navigate through human crowds. Existing approaches usually assume that pedestrians follow a predefined collision avoidance strategy, like social force model (SFM) or optimal reciprocal collision avoidance (ORCA). However, their performances commonly need to be further improved for practical applications, where pedestrians follow multiple different collision avoidance strategies. In this paper, we propose a map-based deep reinforcement learning approach for crowd-aware robot navigation with various pedestrians. We use the sensor map to represent the environmental information around the robot, including its shape and observable appearances of obstacles. We also introduce the pedestrian map that specifies the movements of pedestrians around the robot. By applying both maps as inputs of the neural network, we show that a navigation policy can be trained to better interact with pedestrians following different collision avoidance strategies. We evaluate our approach under multiple scenarios both in the simulator and on an actual robot. The results show that our approach allows the robot to successfully interact with various pedestrians and outperforms compared methods in terms of the success rate.

---

4. Distributed Reinforcement Learning with Self-Play in Parameterized Action Space
   Jun Ma, Shunyi Yao, Guangda Chen, Jiakai Song, and Jianmin Ji.
   SMC 2021
   [Abstract], [Demo], [PDF]
   

   Abstract: Self-play has been shown to be effective to provide a proper training curriculum for a reinforcement learning agent in competitive multi-agent environments without direct supervision. However, its performance is still unstable for problems with sparse rewards, e.g., the scoring task with goalkeeper for robots in RoboCup soccer. It is challenging to solve these tasks in reinforcement learning, especially for those that require combining high-level actions with flexible control. To address these challenges, we introduce a distributed self-play training framework for an extended proximal policy optimization (PPO) algorithm that learns to act in parameterized action space and plays against a group of opponents, i.e., a league. Experiments on the domain of simulated RoboCup soccer show that, the approach is effective and learns more robust policies against various opponents compared to existing reinforcement learning methods. A demonstration video is available online at https://youtu.be/BuL_li1vND4.

---

5. Distributed Non-Communicating Multi-Robot Collision Avoidance via Map-Based Deep Reinforcement Learning
   Guangda Chen, Shunyi Yao, Jun Ma, L. P., Y. C., P. X., Jianmin Ji and Xiaoping Chen.
   Sensors (Volume: 20, Issue: 17 , August, 27, 2020. IF: 3.576)
   [BibTeX], [Abstract], [PMC], [PDF], [Certificate], [YouTube], [bili_1], [bili_2]
   

   @Article{chen2020distributed,
   title = {Distributed Non-Communicating Multi-Robot Collision Avoidance via Map-Based Deep Reinforcement Learning},
   author = {Chen, Guangda and Yao, Shunyi and Ma, Jun and Pan, Lifan and Chen, Yu'an and Xu, Pei and Ji, Jianmin and Chen, Xiaoping},
   journal = {Sensors},
   volume = {20},
   number = {17},
   pages = {4836},
   year = {2020},
   publisher = {Multidisciplinary Digital Publishing Institute},
   doi = {10.3390/s20174836},
   url = {https://www.mdpi.com/1424-8220/20/17/4836}
   }

   Abstract: It is challenging to avoid obstacles safely and efficiently for multiple robots of different shapes in distributed and communication-free scenarios, where robots do not communicate with each other and only sense other robots' positions and obstacles around them. Most existing multi-robot collision avoidance systems either require communication between robots or require expensive movement data of other robots, like velocities, accelerations and paths. In this paper, we propose a map-based deep reinforcement learning approach for multi-robot collision avoidance in a distributed and communication-free environment. We use the egocentric local grid map of a robot to represent the environmental information around it including its shape and observable appearances of other robots and obstacles, which can be easily generated by using multiple sensors or sensor fusion. Then we apply the distributed proximal policy optimization (DPPO) algorithm to train a convolutional neural network that directly maps three frames of egocentric local grid maps and the robot's relative local goal positions into low-level robot control commands. Compared to other methods, the map-based approach is more robust to noisy sensor data, does not require robots' movement data and considers sizes and shapes of related robots, which make it to be more efficient and easier to be deployed to real robots. We first train the neural network in a specified simulator of multiple mobile robots using DPPO, where a multi-stage curriculum learning strategy for multiple scenarios is used to improve the performance. Then we deploy the trained model to real robots to perform collision avoidance in their navigation without tedious parameter tuning. We evaluate the approach with multiple scenarios both in the simulator and on four differential-drive mobile robots in the real world. Both qualitative and quantitative experiments show that our approach is efficient and outperforms existing DRL-based approaches in many indicators. We also conduct ablation studies showing the positive effects of using egocentric grid maps and multi-stage curriculum learning.
   Keywords: multi-robot navigation, distributed collision avoidance, deep reinforcement learning

---

6. Multi-Robot Collision Avoidance with Map-Based Deep Reinforcement Learning
   Shunyi Yao∗, Guangda Chen∗, Lifan Pan, Jun Ma, Jianmin Ji and Xiaoping Chen.
   ICTAI 2020
   [BibTeX], [Abstract], [PDF], [Demo]
   

   @InProceedings{yao2020multi,
   author = {Yao, Shunyi and Chen, Guangda and Pan, Lifan and Ma, Jun and Ji, Jianmin and Chen, Xiaoping},
   booktitle = {Proceedings of the 32th International Conference on Tools with Artificial Intelligence (ICTAI)},
   title = {Multi-Robot Collision Avoidance with Map-based Deep Reinforcement Learning},
   pages = {532--539},
   year = {2020},
   organization = {IEEE}
   }

   Abstract: Multi-robot collision avoidance in a communicationfree environment is one of the key issues for mobile robotics and autonomous driving. In this paper, we propose a mapbased deep reinforcement learning (DRL) approach for collision avoidance of multiple robots, where robots do not communicate with each other and only sense other robots’ positions and the obstacles around them. We use the egocentric grid map of a robot to represent the environmental information around it, which can be easily generated by using multiple sensors or sensor fusion. The learned policy generated from the DRL model directly maps 3 frames of egocentric grid maps and the robot’s relative local goal positions into low-level robot control commands. Compared to other methods, the map-based approach is more robust to noisy sensor data and does not require the expensive movement data of other robots, like velocities, accelerations and paths. We first train a convolutional neural network for the navigation policy in a simulator of multiple mobile robots using proximal policy optimization (PPO), where curriculum learning strategy is used to accelerate the training process. Then we deploy the trained model to real robots to perform collision avoidance in their navigation. We evaluate the approach with various scenarios both in the simulator and on three differential-drive mobile robots in the real world. Both qualitative and quantitative experiments show that our approach is efficient with high success rate. The demonstration video can be found at https://youtu.be/jcLKlEXuFuk.
   Keywords: multi-robots collision avoidance, reinforcement learning, egocentric grid map

---

7. DRQN-based 3D Obstacle Avoidance with a Limited Field of View
   Yu'an Chen, Guangda Chen, Lifan Pan, Jun Ma, Yu Zhang, Yanyong Zhang and Jianmin Ji.
   IROS 2021
   [Abstract], [Demo], [PDF]
   

   Abstract: In this paper, we propose a map-based end-to-end DRL approach for three-dimensional (3D) obstacle avoidance in a partially observed environment, which is applied to achieve autonomous navigation for an indoor mobile robot using a depth camera with a narrow field of view. We first train a neural network with LSTM units in a 3D simulator of mobile robots to approximate the Q-value function in double DRQN. We also use a curriculum learning strategy to accelerate and stabilize the training process. Then we deploy the trained model to a real robot to perform 3D obstacle avoidance in its navigation. We evaluate the proposed approach both in the simulated environment and on a robot in the real world. The experimental results show that the approach is efficient and easy to be deployed, and it performs well for 3D obstacle avoidance with a narrow observation angle, which outperforms other existing DRL-based models by 15.5% on success rate.

---

8. Deep Reinforcement Learning of Map-Based Obstacle Avoidance for Mobile Robot Navigation
   Guangda Chen, Lifan Pan, Yu'an Chen, P. X., Z. W., P. W., Jianmin Ji and Xiaoping Chen.
   SN Computer Science (Volume: 2 , Issue: 6 , August, 18, 2021)
   [BibTeX], [Abstract], [PDF]
   

   @article{chen2021deep,
   title = {Deep Reinforcement Learning of Map-Based Obstacle Avoidance for Mobile Robot Navigation},
   author = {Chen, Guangda and Pan, Lifan and Chen, Yu'an and Xu, Pei and Wang, Zhiqiang and Wu, Peichen and Ji, Jianmin and Chen, Xiaoping},
   journal = {SN Computer Science},
   volume = {2},
   number = {6},
   pages={1--14},
   year = {2021},
   month = {August},
   publisher = {Springer},
   doi = {10.1007/s42979-021-00817-z},
   url = {https://doi.org/10.1007/s42979-021-00817-z}
   }

   Abstract: Autonomous and safe navigation in complex environments without collisions is particularly important for mobile robots. In this paper, we propose an end-to-end deep reinforcement learning method for mobile robot navigation with map-based obstacle avoidance. Using the experience collected in the simulation environment, a convolutional neural network is trained to predict the proper steering operation of the robot based on its egocentric local grid maps, which can accommodate various sensors and fusion algorithms. We use dueling double DQN with prioritized experienced replay technology to update parameters of the network and integrate curriculum learning techniques to enhance its performance. The trained deep neural network is then transferred and executed on a real-world mobile robot to guide it to avoid local obstacles for long-range navigation. The qualitative and quantitative evaluations of the new approach were performed in simulations and real robot experiments. The results show that the end-to-end map-based obstacle avoidance model is easy to deploy, without any fine-tuning, robust to sensor noise, compatible with different sensors, and better than other related DRL-based models in many evaluation indicators.
   Keywords: robot navigation, obstacle avoidance, deep reinforcement learning, grid map

---

9. Robot Navigation with Map-Based Deep Reinforcement Learning
   Guangda Chen, Lifan Pan, Yu'an Chen, P. X., Z. W., P. W., Jianmin Ji and Xiaoping Chen.
   ICNSC 2020, [Best Student Paper Award]
   [BibTeX], [Abstract], [PDF], [arXiv], [Demo], [Slides], [Award]
   

   @InProceedings{chen2020robot,
   title = {Robot Navigation with Map-Based Deep Reinforcement Learning},
   author = {Chen, Guangda and Pan, Lifan and Chen, Yu'an and Xu, Pei and Wang, Zhiqiang and Wu, Peichen and Ji, Jianmin and Chen, Xiaoping},
   year = {2020},
   pages = {1-6},
   address = {Nanjing, China},
   doi = {10.1109/ICNSC48988.2020.9238090},
   organization = {IEEE}
   }

   Abstract: This paper proposes an end-to-end deep reinforcement learning approach for mobile robot navigation with dynamic obstacles avoidance. Using experience collected in a simulation environment, a convolutional neural network (CNN) is trained to predict proper steering actions of a robot from its egocentric local occupancy maps, which accommodate various sensors and fusion algorithms. The trained neural network is then transferred and executed on a real-world mobile robot to guide its local path planning. The new approach is evaluated both qualitatively and quantitatively in simulation and realworld robot experiments. The results show that the map-based end-to-end navigation model is easy to be deployed to a robotic platform, robust to sensor noise and outperforms other existing DRL-based models in many indicators.
   Keywords: robot navigation, obstacle avoidance, reinforcement learning, occupancy map

---

10. Accurate Intrinsic and Extrinsic Calibration of RGB-D Cameras with GP-based Depth Correction
    Guangda Chen, Guowei Cui, Zhongxiao Jin, Feng Wu and Xiaoping Chen.
    IEEE Sensors Journal (Volume: 19 , Issue: 7 , April, 1, 2019. IF: 3.301)
    [BibTeX], [Abstract], [PDF], [PDF_2]

    @Article{chen2019accurate,
    title = {Accurate Intrinsic and Extrinsic Calibration of {RGB}-D Cameras With {GP}-Based Depth Correction},
    author = {Guangda Chen and Guowei Cui and Zhongxiao Jin and Feng Wu and Xiaoping Chen},
    journal = {IEEE Sensors Journal},
    volume = {19},
    number = {7},
    pages = {2685--2694},
    year = 2019,
    month = {apr},
    publisher = {IEEE},
    doi = {10.1109/jsen.2018.2889805},
    url = {https://doi.org/10.1109%2Fjsen.2018.2889805}
    }

    Abstract: In recent years, more and more robots have been equipped with low-cost RGB-D sensors, such as Microsoft Kinect and Intel Realsense, for safe navigation and active interaction with objects and people. In order to obtain more accurate and reliable fused color and depth information (coloured point clouds), not only the intrinsic and extrinsic parameters of color and depth sensor should be precisely calibrated, but also the external corrections of depth measurements are required. In this paper, using motion capture system, we propose a reliable calibration framework that enables the precise estimation of the intrinsic and extrinsic parameters of RGB-D sensors and provide a model-free depth calibration method based on heteroscedastic Gaussian Processes. Compared with the existing depth correction techniques, our method can simultaneously estimate the mean and variance of the depth error at different measurement distances, i.e., the probability distribution of the depth error relative to the measured distance, which is essential in the state estimation problems. To verify the effectiveness of our approach, we conduct a thorough qualitative and quantitative analysis of the major steps of our calibration method, and compare our experimental results with other related work. Furthermore, we demonstrate an experiment about the overall improvement of visual SLAM with a Kinect device calibrated by our calibration technique.
    Keywords: RGB-D cameras, calibration, motion capture system

---

11. 一种分布式多机器人的导航方法[P]
    陈广大, 姚舜一, 吉建民.
    中国专利：CN112304314A, 2021-02-02.

* These authors contributed equally to the work.

Conference reviewer : IROS 2019, ICRA 2021, SMC 2021, ICRA 2022, IROS 2022

Journal reviewer : IEEE Access, Complexity, Applied Artificial Intelligence

Some links : arxivdaily.cs.RO, USTC GitLab, Song-Jia Yi's ORCID