Center for Intelligent Bionic

Introduction

Center for Intelligent and Biomimetic Systems was officially established in September 2006. Academician, Professor, Yangsheng Xu is the director. Three elements, perception, cognition, and action of robotic research, are explored as the core technology and key scientific problem of our research center. Integration features from machinery, information, biomimetic, medicine or other interdisciplinary are guided by advanced integration technology, dedicated to domestic service robots, medical and rehabilitation robotics, microrobots, and intelligent systems for aspects of cutting-edge research and industrialization.

 

Our Center led the construction of the “Guangdong Provincial Key Laboratory of Robotics and Intelligent System”, “Guangdong Engineering Technology Research Center of Robotics and Intelligent System”,  “Shenzhen Intelligent Robot Engineering Laboratory ”, establishing China's first robot association “Shenzhen Robot Association”, organizing the establishment of the “Guangdong Provincial Robotic Industrial Technology Innovation Alliance”.

 

Group photo of the Center in 2016

Main Research Areas

1.Exoskeleton robot

The quantity of people suffering from walking disturbances shows a trend of rising in recent years. Spinal cord injuries may lead to hemiplegic or paraplegia. And a large number of people have difficulty walking due to old age. However, rehabilitation equipments in hospitals still have several deficiencies. First of all, a lot of rehabilitation equipments need collaboration of multiple caregivers, which is laborious and time consuming; Second, in the process of using, a lot of equipments lack of tracking patients' physical and biological conditions; Moreover, some complex equipments require quite a long time to prepare, leading to low efficiency, and the maintenance cost is very high; Finally, most of the existing rehabilitation solutions are subject to environmental constraints. The lower limb Exoskeleton robot has many advantages, such as, the caregiver can take care of more than one patients at the same time. The robot can record the physiological state of patients simultaneously. And it is easy to wear and take off, and easy to adjust to local conditions.

 

2.Climbing robot for nuclear power plant.

Steam generator, as interactive equipment for the reactor, is the pivot for the first and secondary loops. The climbing robot is developed to diagnose the steam generator. The robot is modularly designed. It could climb and rotate freely on the inner wall of the secondary loop. A comprehensive visual investigation to central tunnel, outer corridor of the secondary loop, upper side of board and the crack and surface of the heat transfer tube could be achieved by the actuator located in robot’s end. Real time video and image could be transmitted to remote controller. The robot has features as stiff adsorption, free climbing, remote controllability, auto-inspection, video inspection in narrow space, visual simulation. The comprehensive detection to the secondary loop of steam generator could be achieved from these features. The robot passed nuclear radiation experiment and is approved by institutes as the society of mechanical engineering of Guangdong province and quality inspection institute of Shenzhen.

steam generator model           system layout and component of the climbing robot

 

the main part of climbing robot

 

3. Inspection robot for base station and engine room

Focusing on unmanned base station in remote rural area, the Inspection robot for base station and engine room is a robot with capacity of remote control and auto pilot inspection, satisfying the actual demand of a certain company. A variety of functions could be achieved by this robot, including auto-inspection of the equipment in base station and engine room, live screen return, infrared remote control of electrical equipment, equipment temperature detection, environment temperature and moisture inspection, vocal intercommunication, auto-pilot. The robot is composed of moveable vehicle, adjustable PTZ and charging pile. The performance of the robot is as below:

The robot could move steadly with even velocity on slope within 10°. The velocity is limited within 0.1~0.2m/s.

The robot could move, turn around and change direction in a corridor whose width is 1m.

The robot could cross slim barrier with width of 3cm and height of 1cm.

Ultrasonic distance detectors and infrared signal receivers are distributed evenly on the moveable vehicle, enable the robot to be auto collision avoidance and auto pilot.

The PTZ that the robot carries has three degree of freedom, that is, stretch along vertical axis, rotation around vertical axis and rotation around horizontal axis.

Fig. X Component of the Hub robot system 

 

4.Research on Key Technology and Applications of Robot Intelligent Control Based on Learning Human Actions

Although the fundamental research and commercialization on robotics have been greatly developed for several years, the intelligent decision making abilities of robotics on planning and control are still falling far behind human beings, no matter on manufacturing, helping the aged, medical care and relief, search and rescue, etc., which is a problem restricting the development of robotics for a long period of time. To tackle this problem, we focus on in-depth theoretical studies on three aspects based on the learning and imitation of human actions, including high-level task planning, complex trajectory tracking and position/force control.

Specifically, the main research contents include three aspects: (1) Research on high-level planning methods of how to combine the operating primitives through learning from human demonstrations; (2) research on the intelligent tracking methods of anthropomorphic complicated paths, and ensure the convergence on a specific working point; (3) research on the applicability of common robotic compliance control methods (such as impedance control, position/force control etc.), and develop new intelligent compliance control methods. Furthermore, we build a variety of experimental platforms, such as industrial robots, rehabilitation robots etc., to verify the proposed theory and methods of robotic anthropomorphic behavior control studies.

Application Scenarios

 

5.Low-cost robot SLAM

This project mainly studies the global map construction technology and trajectory planning and trajectory tracking technology of low cost indoor robot under complex scene. It contents three aspects: (1) Solving the contradiction between low-cost computing unit, low-precision sensor and high-precision positioning, self-construction; (2) trajectory tracking control; (3) proposed the algorithm of path planning and obstacle avoidance based on improved heuristic algorithm.

 

6.Service robot development

This project mainly carries on the research and the modular development of the core technology of the robot core, including: indoor navigation algorithm and module, human body detection and motion recognition algorithm and module, humanoid robot coordination control module and so on.

 

7.Magnetically actuated helical microswimmers: visual servo control

Helical microswimmers capable of propulsion at low Reynolds numbers have been proposed for numerous applications, ranging from in vitro tasks on lab-on-a-chip (e.g. transporting and sorting micro objects; mechanical components micro assembly...) to in vivo applications for minimally invasive medicine (e.g. targeted drug delivery; brachytherapy; hyperthermia...), due to their micro sizes and accessibility to tiny and clustered environments.

Several kinds of artificial helical microswimmers actuated by magnetic fields have been proposed by researchers. However, they are primarily open-loop controlled, due to the complexity of the control of the magnetic field actuating the helical propulsion, and the limited number of feedback options processing the required precision. Therefore, one of the main objectives of my research is to investigate methods of closed-loop control of a magnetic-actuated helical swimmer at low Reynolds number. A planar path following control of a helical swimmer with 3D steering is realized using visual servo methods based on the real-time visual tracking with a stereo vision system.

 

 

8. Rescue robot

Joint intelligent rescue robot, as a handy helper for the rescue crews, could help the rescue crew search and save the suvivors in perilous and sophisticated circumstance, shortening rescue time and ensuring safety of the rescue crews to the great length.

The robot could carry a number of small robots. The multi-combination of these small robot, detemined by specific environment, could provide a variety of solution to multi-task, the environment adaptability of the robot enhanced on a big scale.

In data collection module, environment detection techniques like ultrasonic ranging, infrared ranging, temperature detection, life signal detection, gas monitoring, electronic compass, GPS, sound positioning, digital encoding discs, video capture are integrated by multi-sensor fusion, constuting a integral digital information system. From the data gathered by sensors, the system could not only receive real time position and terrain information, but also provide precious data for exploration and rescue.

 

9.Pet robot Nabao

Pet robot Nabao is a nexus of bio-inspired robot, muti-touch, interactive projection. Nabao is composed of movement unit, touching unit and wireless interactive unit. The servo motor that movement unit adopted has modular middle shaft, giving Nabao superior maneuverability and compact structure. The touching units locate at all main bodies on Nabao, enable Nabao to receive touching in different places and receive feedback. User could feed Nabao virtually by multi-touch platform. Nabao would bark joyfully and dance with the beat of music. There locates three-axis accelerometer, enable Nabao to perceive its current location and orientation. Nabao could stand in right position automatically when it is pulled down. The anxious emotion could be sensed by its unpleasant expression. The vivid movement and bio-inspired function transcend the normal concept of robot.

Nabao is oriented to children, old people and white-collared worker. Children could feed, play and dance with Nabao. Nabao could disperse the atmosphere of loneliness for old people. For white-collared workers, Nabao could alleviate stress and anxiety for its playfulness.

The pet robot Nabao, sophisticated electronic pet, could be virtually a pet for domestic leisure and amusement.

 

10. Object recognition and 3D pose estimation based on deep learning technology

Object recognition and pose estimation is one of the basic problems in the field of artificial intelligence and computer vision. This research proposed an improved approach based on a multi-stage feature learning network to identify objects as well as their 3D orientation. This technology can be widely used in the industry of manufacturing, processing and assembling.

 

 

11.The non-contact four-wheel positioning system for commercial vehicles

This project aimed to develop an intelligent non-contact four-wheel positioning system for commercial vehicle which can be used to calibrate the four-wheel positioning parameters to meet the requirements of automobile factories as well as garages. The main technology includes feature extraction and matching algorithm, the wheel pose estimation algorithm based on line structured lights and stereo vision system, and the real time wheel recognition and positioning technology.

 

12.The intelligent vision system for automatic screw machine for flexible manufacturing

The project proposed an intelligent recognition technology to identify the screw holes’ positions in the workpiece. Cooperated with the Mechanical mechanism of the screw machine, the system can achieve automatic screw tightening operation with high accuracy.

 

13.Heterogeneous sensing system-based Unmanned Aerial Vehicle navigation

Monocular Simultaneous Localization and Mapping (SLAM), due to the low weight of monocular camera, is an area of popular research and promotes countless applications of micro Unmanned Aerial Vehicles (UAVs), especially in some GPS-denied indoor environments. However, the motion of UAVs is often faster and more complex than that of ground-based robots. It would also lead to error accumulation if we calculate the trajectory only through the ego-motion. The fusion of visual and inertial sensors has proved to be popular in robot navigation and, in particular, UAVs navigation due to its flexibility in weight, low power consumption and cost.

UAV indoor positioning with data fusion

Researchers

  • WU Xinyu

    Title:Professor
    Email:xy.wu@siat.ac.cn
    Areas of Interest:Robot, Intelligent System, Computer Vision

  • WANG Can

    Title:Associate Professor
    Email:can.wang@siat.ac.cn
    Areas of Interest:Networked robots, Exoskeleton robots

  • LIANG Guoyuan

    Title:Associate Professor
    Email:gy.liang@siat.ac.cn
    Areas of Interest:Image processing, Computer vision

  • XU Tiantian

    Title:Associate Professor
    Email:tt.xu@siat.ac.cn
    Areas of Interest:Microrobotos, control, path following, visual servoing

  • GUO Shifeng

    Title:Professor
    Email:sf.guo@siat.ac.cn
    Areas of Interest:Sensors and sensor networks, smart materials and structures, ultrasound based non-destructive testing (NDT)

  • ZHANG Yanhui

    Title:Associate Professor
    Email:zhangyh@siat.ac.cn
    Areas of Interest:Intelligent detection method

  • ZHOU Yimin

    Title:Associate Professor
    Email:ym.zhou@siat.ac.cn
    Areas of Interest:Control Theory and Engineering

  • CHEN Chunjie

    Title:Senior Engineer
    Email:cj.chen@siat.ac.cn
    Areas of Interest:Exoskeleton robot, service robots and Robotic


  • 1