Kinematic identification of Tripteron robot is one of the purposes of this thesis. Then precise position control could be achieved using identified kinematic parameters of Tripteron robot. For determining the position of end-effector, a special 3-D camera (named Kinect) will be used and image processing algorithms will be applied. In addition, the camera will be used for position control and obstacle avoidance. Finally, the robot should be able to accurately pursue any desired trajectory.

The last step of this thesis is involved with design and development of a user-friend GUI (Graphical User Interface) so that user can simply interact with robot.

 

In Dynamical Control & Identification of Tripteron, as a decoupled parallel robot performing 3 translational motion, the main goal is to control the robot by using the torque mode control of the actuators. As a joint work, the dynamic model of the robot will be obtained by MohammadAli Bagherian. In order to reach more accurate control performance, the team will perform system identification techniques on the dynamic model of the robot. This procedure will involve both White-Box Identification (based on provided model from theoretical concepts) and Black-Box Identification (Data Based). The main purpose to perform the identification is to reach the most accurate possible model of system on the presence of uncertainties which are neglected in theoretical dynamics model such as friction and backlash in the joints of the robot. The dynamical control of system would be performed from simple and classical PID controller to more complex ones to reach a stable and efficiency control of the system.

 

Nao humanoid robot v4 is currently under study in order to explore various topics. NAO is a position-based control humanoid which is one of the main difficulties. System identification is a key point to overcome to these difficulties, and make useful some torque-based control laws. Currently, system identification based on frequency analysis and model based methods are studying which will introduce a transformation for torque to position commands.

Moreover, identification of robot parameters such as masses and inertia tensors, and the motor parameters of each joint.

 

This project is concentrated on identification-based control methods of pneumatic 6-DOF Gough-Stewart platform. Gough-Stewart platform is a type of parallel robot, having a stationary base platform upon which a moving platform is supported by six links.

For the first step, the control of the Gough-Stewart platform is based on data provided by the pressure sensors. Also local and cascade control loops are suggested. Then a rotation sensor is installed on the end-effector in order to have close-loop control on the end-effector orientation.

Moreover, the IKP (Inverse Kinematic Problem) and FKP (Forward Kinematic Problem) is used by applying an inner and outer control loop. The main objective in this project consists in using all sensors, such as pressure sensor, proportional, orientation sensors by applying data fusion concepts in order to improve the performance of the robot. In addition, for the sake of a better control, vision techniques are applied to calibrate the robot, and a user friendly Graphical User Interface (GUI) is designed.
It should be noted that due to the non-linearity behavior of the mechanisms different control techniques such as sliding mode, adaptive, evolving and STR (Self Tuning Regulator) control methods combined by identification techniques are used.