Design and implementation of traffic sign and obstacle detection in a self-driving car using SURF detector and Brute force matcher

One of the most amazing innovations of smart city traffic management and control is the upsurging technological advancements in the field of automated vehicles. This paper focuses on implementation of the core feature of a self-driving car which is traffic sign and obstacle detection. This was made possible through Feature matching Image processing algorithm using a combination of SURF detector, Brute force matcher and RANSAC. This Feature matching algorithm was preferred as they are least dependent on lighting conditions and is rotation or tilt invariant. This algorithm was tested in realtime and has proved to show least errors even at high speeds. Along with this algorithm, a highly accurate distance calculation method was proposed by modifying basic optics principles to give accurate distance values from the object. Hardware was implemented and tested on a 2 wheel drive robot accompanied by a Raspberry pi and MCU circuit.