Human & Robot Interaction Lab. (TaarLab)

A+ A A-
People Involved
   Mojtaba Yazdani, Roya Sabbagh Novin

One of the challenging subjects in industrial serial manipulators context is controlling them to continue their tasks in the case of actuator faults. This is usually done considering minimum velocity jump in their end-effector and healthy joints.

The method suggested in this research, utilizes Model Predictive Control (MPC) in order to control the manipulator for two scenarios; the first one is when the joint trajectories are specified offline and the controller has to modify them to reduce the jump of the end-effector, and the other one is online motion planning, in which the jump of the end-effector is zero and the objective function is defined to minimize the jump in the joint trajectories. In the presence of actuator fault, using optimization techniques, the optimal solution is determined which minimizes the velocity jumps of healthy joints.

For experimental tests, a 4-DOF planar serial manipulator named “Taarm” (Taarlab’s Arm), is used. Since the desired task is a 2-DOF task, the manipulator will remain redundant when there is only one actuator fault. Moreover, E-puck mobile robots are used as dynamic obstacles.

It should be mentioned that all algorithms in simulation and experimental tests are developed in C++ with QT Creator IDE in Linux and the optimization problems are solved via Gorubi optimization package which is one of the best real-time optimization packages.

taarm1 smdiag1

Last Updated on Sunday, 08 February 2015 13:43

Hits: 570

People Involved
   Roya Sabbagh Novin, Mojtaba Yazdani, Amirhossein Karimi

This project focuses on the optimal path planning of redundant planar serial robots, while avoiding obstacles within its workspace. A synergy-based algorithm between convex optimization, disjunctive programming and receding horizon is employed to the end of achieving advantages such as finding the global optimum solution and low computational time. For the purpose of the problem, different cost functions can be considered, including among others, transition time, energy usage or path length. In addition, kinematic and dynamic relations of the robots under study are expressed as constraints of the problem and since they are non-convex functions, they are approximated by convex constraints.

So far, the proposed algorithm is simulated for 3-DOF and 4-DOF redundant planar serial robots in the presence of static and dynamic obstacles using CVX package in MATLAB. Now it is being tested on a 4-DOF planar serial manipulator developed in TaarLab named as “Taarm”.

Moreover, the same framework will be used for path planning of parallel robots by some minor modifications.

smdiag2 convex 

Last Updated on Sunday, 08 February 2015 13:46

Hits: 604

People Involved
   Mojtaba Yazdani, Roya Sabbagh Novin

In order to have an experimental test platform for our optimal motion planning and fault tolerant control projects, a 4-DOF planar serial manipulator is developed. This manipulator contains four AX-12 Dynamixel actuators with good performance in position control which are connected to PC via a USB2Dynamixel card.
It should be mentioned that all experimental test algorithms should be developed in C++ at Linux to have a real-time interaction with this manipulator.


Last Updated on Sunday, 08 February 2015 13:55

Hits: 516

People Involved
   Behzad Mehrafrooz

In this investigation serial manipulators are studied via Open dynamics engine.

The first part of study concentrates on planar robots. In the next stage spatial robots are investigated. Also In this work a model for obtaining dynamics equation of an n-linkage robot is presented. We also linked SolidWorks CAD files with MATLAB to demonstrate our result in a GUI (graphical User Interfaces).


Last Updated on Sunday, 08 February 2015 13:53

Hits: 725

TaarLab 2014, All Rights Reserved.
Designed, Developed and Powered by: Payam Ghasemi & Nima Karbasi