A range security system for vehicle navigation

The contribution of this research is the development of a security system the purpose of which is to assist the teleoperator of a mobile robot by providing him pertinent information about the robot working space. In this aim, the range image based perception system must be able to detect and localize, as fast as possible, obstacles (nontraversable regions) in the scene during robot's motion. In this paper we present an accelerated strategy of obstacle detection using the knowledge of objects contained in previous images and their predicted positions in the new image. An iterative procedure based on a Kalman filter is used to improve robot's state knowledge. Then, an image segmentation takes place according to security and emergency criteria. The range image is divided in priority areas to reduce obstacle research time. Some results obtained through experimentation of this method on synthetic images are developed in this paper.