The focus of this thesis is on active safety, and as such there is a wide range of functions that could be developed, the following were chosen.
A Lane Detection algorithm is developed, described in section 2.2. This function uses image analysis and GPS data to find the lane in which the vehicle is traveling by implementing a Hough transform in near field and a second degree polynomial fitting for the far field.
Lane Departure Warning, LDW, is a common feature in modern cars. With knowledge regarding where in the image the vehicle is traveling, it is also possible to detect when the vehicle is about to leave its lane. This functionality is developed in this thesis as well, described in section 3.5.
As a mean to help the driver Object Detection was developed, as described in section 2.3. It localizes objects ahead of the vehicle in the traveling lane. With information about the detected objects, different warnings or safety functions such as applying the brake or an audio alert can be developed as well.
To test the mobility and agility of the complete model it is ported to a Raspberry Pi 3 and run in full scale using a Logitech HD camera as input.