Humanoid Robot: Throw Ball At A Target

Bhavishya Mittal
Pratibha Prajapati
Supervisor: Dr. Amitabha Mukerjee

Abstract:

Our project aims to identify known target through ALDEBARAN NAO and then to throw an object towards it. In our project we first detect the known target using Image Processing techniques, then we compute the position of that object and finally we perform the throwing action.
The throwing object ability make a humanoid to manipulate the target or object outside the movable reach of the robot. In the throwing action the robot motion is quick and dynamic. There are many ways to perform this action and this makes the inverse kinematics very non-trivial. Also throwing action requires precision between the timings of different part. This project proposes the linear approximation model and feedback learning based approach to throw objects towards target.

Documentation:

PROPOSAL    PRESENTATION    POSTER    FINAL REPORT    CODES

REFERENCES:

  1. Miyashita, Hideyuki, Tasuku Yamawaki, and Masahito Yashima. ''Control for throwing manipulation by one joint robot.'' Robotics and Automation, 2009. ICRA'09. IEEE International Conference on IEEE, 2009.
  2. Aboaf, Eric W., C. G. Atkeson, and D. J. Reinkensmeyer. ''Task- level robot learning.'' Robotics and Automation, 1988. Proceedings., 1988 IEEE International Conference on IEEE, 1988.
  3. Kober, Jens, Katharina Muelling, and Jan Peters. ''Learning throwing and catching skills.'' Intelligent Robots and Systems (IROS), 2012 IEEE/RSJ International Conference on IEEE, 2012.
  4. Master’s Thesis of Tomàs Gonzàlez Sànchez,Dep. of Computer Science and Mathematics, Universitat Rovira I Virgili, September 2009, 64-82
  5. Lovish, Rahul IIT Kanpur, ''Detection(Partial)/Kicking the Ball with Aldebran Nao'', 2013