Sunday 30 April 2017

Machine Vision Module (MVM) for small robots 2 - Integration

April 2017
Completed the integration of the MVM (Version 0.9) into a small two-wheels self balancing robot (the V2T1).  The MVM is powered from the robot's battery, and communicates with the robot's controller via serial port, reporting the MVM's algorithm outputs after every image frame is analyzed.  For instance if we are running an algorithm to detect the brightest object in the scene, the coordinate of the brightest pixel will be send to the robot's controller after each image frame is captured.  Then robot's controller can then take the appropriate action based on this information.  More details of the MVM can be found here.  A video below show's the preliminary result:

August 2017
A minor update to the image processing algorithm.  Now the robot can use the MVM (Version 0.9) to look out for dark lines.  Here the video is showing the robot 'patrolling' in a region bounded by dark tapes.

May 2018
Integrated MVM (Version 1.0) to two-wheels self-balancing robot platform V2T2, with a pair of 4 degree-of-freedom arms.