Investigation into Image Processing Techniques Capable of Distinguishing
Inclusive Dates: 07/08/09 – 11/08/09
Background - Demining machines are currently used throughout the world to remove anti-personnel mines by tilling the soil. These durable machines are designed to withstand ordnance detonation and are used after anti-tank mines have been cleared from the mine field. Current techniques for operating the machines depend upon high resolution Global Positioning System (GPS) tracking and are therefore expensive. Tele-operations are often limited as it is difficult to drive the machine effectively in real time from a safe distance while maintaining the required overlap of the demining machine and the cleared terrain. The need to increase the efficiency of the demining process is great, as currently mines are laid at a rate 25 times faster than they are being cleared.
Approach - This research investigated and developed image processing techniques and algorithms capable of detecting cleared versus uncleared paths from a camera pointing directly in front of the demining machine during mine clearing operations. The approach taken involved three core tasks:
The algorithms were initially developed in a National Instruments LabVIEW environment and then ported to C++ for testing with vehicle control and path planning algorithms previously developed under the Mobile Autonomous Robotics Technology Initiative (MARTI).
Accomplishments - The team successfully developed algorithms capable of detecting cleared versus uncleared paths from the video provided. The investigators have used these algorithms to provide a visual overlay of the video showing the overlap of the machine to the previously uncleared path and providing feedback for an operator (or vehicle control algorithm) to help control steering. This research will greatly improve the ability to tele-operate the machine leading to the ultimate goal of machine automation. The benefits of this research extend well beyond the demining industry to include agriculture/farming, construction, and mining.