INTER VEHICLE DISTANCE AND RELATIVE VELOCITY ESTIMATION

Abstract

The notion of fully autonomous personal vehicles has shifted from a dream to inevitabily in recent years, bringing a host of technical and ethical challenges. Clearly, the technology in these vehicles is mission-critical: any failure could have disastrous consequences. The conditions in which autonomous vehicles might operate are incredibly diverse, including the weather, scene and time of day; the on-board systems must therefore be robust to many possible environments. The technology behind autonomous driving is a very active area of research. One of the most critical pieces of information in an autonomous vehicle is the speed of vehicles near by our own vehicle; any miscalculation can affect both trajectory calculation and over or under braking, potentially resulting in unnecessary accidents. This project aims to discover whether camera footage, using no other sensor input can be used for speed calculation of vehicles nearby our own vehicle. Such a standalone system is more robust to failure of individual components in other areas of the vehicle. To predict speed from video, some features of vehicle like wheels centre, license plate etc can be used or we can make use of road scene features like lane marking etc.