Parallel robots

Parallel robots are closed-loop kinematic mechanism in which the end-effector is linked to the base by several independent kinematic chains. It is widely recognized that parallel robots have some potential advantages such as high speed, accuracy, stiffness, payload, and so on. Due to the closed-loop mechanism, the kinematics and dynamics of parallel robots are tightly coupled and highly nonlinear. Thus the coordination manipulation of the end-effector with multiple kinematic chains makes the motion control of parallel robots a work full of challenge. In recent ten years, we have been studying the motion control of robots by using the theory methods of modeling, control, planning and optimization. We have proposed and implemented some innovative ideas such as kinematic self-calibration, optimal dynamic identification, nonlinear friction compensation, nonlinear PD+ controllers, adaptive controller, acceleration feedback controller, and coordination controller.


Humanoid robots

Humanoid robots have humanoid appearance and could communicate with us friendly in our living environment. Biped walking is the most remarkable feature of humanoid robots, thus gait planning is essential. One of our work is realizing a gait planning algorithm based on the linear coupled oscillator model. To reduce the parameter tuning time and simplify the gait planning algorithm, a parameter choosing algorithm is designed by the optimization method. Another research work is the balance control of motion, to realize some complex motion tasks such as crossing uneven ground or climbing stairs.  Moreover, in the area of human-robot interaction, it is important to detect and track people in human surroundings. Our aim is to realize the person following by using the humanoid robot-Nao. According to the images captured by the camera of Nao, the color and contour feature of the person can be detected. And then robot Nao can track the motion of the person.



Navigation is the base to complete the task smoothly for mobile robots. To navigate in an unknown environment, a robot need to know where it is. A SLAM (Simultaneous Localization and Mapping) system can solve this problem by building a map with its sensors and localizing the robot at the same time. Now we are using RGB-D cameras to do research on RGB-D SLAM. RGB-D SLAM algorithms include two parts: frontend and backend. While the frontend extracts spatial relations between individual observations, the backend optimizes the poses of these observations to reduce nonlinear errors. Recently, we are studying the frontend of RGB-D SLAM algorithm, in order to build a RGB-D SLAM system which has a high speed to build a 3D map and localize the robot. And then we will continue to study the backend of RGB-D SLAM algorithm.


Object Recognition and Localization

In many robot applications, interaction with outer world is the final aim of robot tasks. Before robots manipulate objects in world under control, it is basic to recognize and localize objects properly. Using 3D vision method to help robot do more complex jobs has been researched for a few years. Geometry features can be extracted from 3D data, which have enhanced the power of vision. We aim to improve some algorithms of 3D vision and popularize this vision method to traditional industrial robots.

About Us | Contact Us| ©2015-2016 V-Climber