Paper Title
Distance Estimation In Automation Systems Using Feature Based Techniques
Abstract
Automation has become an integral part of most of the industries. The dependency on robots and automation
system for every application necessitates the use of efficient motion estimators. This project proposes the development of a
visual odometer. A visual odometer differs from its conventional version in the sense it works with the processing of the
images captured by a camera. This means that by processing the images captured by an on-board camera, the displacement
of the system or a vehicle can be calculated. A downward facing camera working at high speed captures several pictures of
the surface below the device every second and sends these pictures to the processor. The processor performs feature
detection and pattern matching on the images taken on the camera. As the device moves, the shifts in the patterns of the
captured images are observed and the amount of shift taking place in both axes are calculated. This can help in calculating
the total distance travelled, the velocity and acceleration of the vehicle. This paper presents a study of various feature
matching techniques and an algorithm for implementation of a visual odometer on a level one robot used as a hardware
prototype.
Keywords— Automation systems, odometer, GPS, SIFT, SURF.