We are very proud to be working with a prestigious Korean University with such a strong focus on developing new technologies for the future. DGIST is a totally research focused university that has a remit to develop technologies that can be used to advance Korea's high-tech industries.
This advanced model based development project was to develop a robust video processing and vision recognition algorithm to determine the distance and angle of the vehicle relative to the right-hand lane markings.
From the initial requirements it was decided that the project shall use Matlab and Simulink and the Image Processing toolbox from Mathworks. This greatly speeded up development and enabled many more algorithm designs to be tested throughout the project compared to other options.
The final solution implemented a range of techniques including Hough Transforms and a Kalman Filter. The algorithm switched between the front and rear camera depending upon the confidence of the results and gave very robust results even in urban driving with strong shadows and heavy traffic.