Research

Projects

• Ergonomic Robotic Escort Inferring Human Intention and Utilizing Motion Patterns
  For robotic escort, a robot moves in front of a human and complies with the human aim without direct communication. This work introduces motion patterns to both infer the human intention and control the robot. A certain level of autonomy is provided to the robot, which leads to capable ability of running in cluttered environment and easier operating.
  
  
• Autonomous Map Refinement
  Many SLAM techniques focus on global accuracy and static environment. This work plans the path to revisit the environment and collects more local information as well as the environmental changes. The path planning algorithm is realeased as a ROS Package eulerian_contour_path_planner.
  
  
• Intention Inference and Motion Prediction
  Associates the recurring behaviors of a moving agent with intentions. Infers the intention and models the motion pattern. Utilizes the intention and motion pattern to improve the state estimation.
  Since an agent's intention is usually related to environmental objects. The relations among the target and objects are extracted from past observations for improved intention inference.
  
• Data Annotation Platform and Machine Learning Techniques for Robotics
  The platform targets the annotation of robotics data and uses the current machine learning techniques to aid human annotation.
  Docker containers with ROS wrappers of different techniques (object detection of image and point cloud, semantic segmentation, 3D pose estimation, etc.) are created. The techniques aim for faster development and deployment of robotic applications.
  
• Mohamed Bin Zayed International Robotics Challenge
  Lead the autonomy development for aerial vehicles. In charge of the modules of perception and planning. Designed the cooperation strategy of aerial vehicles and ground vehicles.
  
  
• Experimental Resilience Assessment of an Open-source Driving Agent
  Presented a Systems-Theoretic Process Analysis (STPA) based fault injection framework to assess the resilience of an open-source driving agent -- Openpilot. Extensive numerical experiments are performed by simulating different environmental conditions and faults of sensors to evaluate the pros and cons.
  
  
• Reinforcement Learning for Imitating Human in Autonomous Driving
  Designed an approach of apprenticeship learning based on deep deterministic policy gradient (DDPG) for autonomous driving. This work takes advantage of the human experience to shorten the learning time towards a satisfying autonomous agent.
  
  
• Motion Planning and Safety of Autonomous Robotic Surgery
  Modeled Raven II surgical robots and the working environment using Gazebo. Designed the autonomy pipeline for the tasks of medical debridement. Tested different motion planning algorithms for collision avoidance in autonomous robot surgery.
  
  
• Real-Time Operating System on ARM
  Implemented real-time operating system features on an ARM microcontroller, such as thread management, time management, and priority scheduler.
  Background music of super mario is played through the buzz of the device~
  
  
• Force Interaction of Teleoperation Robots
  Designed a safe force interaction approach for the teleoperation robot. Normal force exerted by the human is transmitted to the robot side while large force is conditionally decreased.

Publications

• Yongming Qin, Makoto Kumon, and Tomonari Furukawa. Robotic Escorting Describing Human Behavior as a Model of Intention and Motion Pattern. Journal of Field Robotics (under review)


• William Smith, Yongming Qin, Tomonari Furukawa, and Gamini Dissanayake. Autonomous Robotic Map Refinement for Targeted Resolution and Local Accuracy. The 2022 IEEE International Symposium on Safety, Security, and Rescue Robotics (SSRR)


• Yongming Qin, Makoto Kumon, and Tomonari Furukawa. Shared Autonomy for Robotic Escort incorporating Human Intention and Motion Patterns of Demonstration. International Conference of Robotics and Automation (ICRA) Workshop 2022


• Yongming Qin, Makoto Kumon, and Tomonari Furukawa. A Data-Driven Multiple Model Framework for Intention Estimation. International Conference of Robotics and Automation (ICRA) 2022


• Yongming Qin, Makoto Kumon, and Tomonari Furukawa. Estimation of a human-maneuvered target incorporating human intention. Sensors, 2021


• Rubaiyat, Abu Hasnat Mohammad, Yongming Qin, and Homa Alemzadeh. Experimental resilience assessment of an open-source driving agent[C]. IEEE Pacific Rim International Symposium on Dependable Computing, 2018


• Yuanqing Xia, Yongming Qin, Dihua Zhai, et al. Further results on cloud control systems[J]. Science China: Information Sciences, 2016, 59(7): 073201


• Yongming Qin, Yuanqing Xia. Control of bilateral teleoperation with time delay considering operator force sending[C]. IEEE 2015 34th China Control Conference, 2015: 6697-6702

Ongoing
• Yongming Qin, William Smith, and Tomonari Furukawa. Solving Euler Circuit Problem for Coverage Path Planning of Object Contours. Robotics and Automation Letters.


• Yongming Qin, Makoto Kumon, and Tomonari Furukawa. Ergonomic Robotic Escort Inferring Human Intention and Learning Motion Patterns from Demonstrations.