Biorobotics Laboratory BioRob

Print

Programming the robotic dog AIBO

The aim of my work is to make AIBO able to detect a mark on the floor using his camera. Then it will place his nearest paw on the detected mark.

In this work, there are three main objectives. The first one is to develop a robust mark detection algorithm by using image processing techniques. The second objective is to create a 2D to 3D mapping from the position in the picture to the position in the floor. Finally the third objective is to develop an inverse kinematics based leg controller. The entry of this controller is the previously computed position of the mark in AIBO reference frame, and the outputs are the leg joint angles that make the paw reach the mark.

Solutions
  • Mark detection: By knowing in advance the color of the mark, we can compute its position in the picture by using a virtual mask moving on all pixels: The mean color inside the mask is calculated to find the mark position in the picture.
  • 2D to 3D mapping: To construct 2D to 3D correspondence matrices, a supervisor in webots makes move a small square on the foor in front of the AIBO camera such that each pixel of the picture 'see' successively the small moving mark. Then, the supervisor does the link pixel(i,j)/ 3D position on the floor.
  • Placing the leg on the mark: The legs joint values are computed by using inverse kinematics.

    Contact persons


    Supervisor: Sarah DEGALLIER
    Responsible Professor: Auke Jan IJSPEERT

    Downloads

    Links