An efficient obstacle awareness application for Android mobile devices

Recent developments in the mobile devices have made image processing on the go much more feasible. In this paper, we propose a monocular approach that can be deployed on smart phones and tablets and we evaluate its performance and the potential for further development of driving assistance system using mobile devices. This monocular approach relies on the main camera of the mobile device to observe the environment. Using the camera calibration parameters, an inverse perspective mapping of the scene is created, and then segmented into obstacle and free areas. The obstacles are then identified by polar and radial histogram processing. From real world experiments, we found that through modern mobile devices we can accurately detect obstacles in real time using a single camera. Developing driving assistance software is possible with the aid of such devices in urban roads, but also on highways.