Stationary obstacle detection using pyramidal lucas kanade optical flow

Abstract

The obstacle detection system could be performed using Pyramidal Lucas Kanade Optical Flow calculation. The calculation is using greyscale images obtained from a camera or video files as input. The results from optical flow calculation then processed into Time to Contact (TTC) value that is used to estimate the distance of the robot to the obstacle. This paper proposed an idea to combine Pyramidal Lucas Kanade Optical Flow, the Region of Interest and Time History properties which can be used to detect an obstacle. The system can display the direction indicator as a response to avoid an obstacle. The results indicate that the system can detect the textured stationary obstacle in environments with light intensity between 22 lx and 400 lx. The average of total execution time required by the system for each frame to be processed is 31 ms.