Feature based indoor mapping using a bat-type UWB radar

A system for building a feature-based map of an unknown indoor environment is proposed. We consider emergency scenarios where smoke and dust block the vision so that camera-based systems and laser range finders are not operable. In this kind of situation, ultra-wideband (UWB) radar is a good alternative. We use a bat-type UWB radar system composed of one transmitter and two receivers arranged in a fixed linear array. Time-of-flight measurements are taken between the transmitter and the receivers. The environment is reconstructed from impulse responses by extracting information on different features like walls, edges and corners. These features can be used as landmarks to navigate. The solution we present works by dividing the problem into three different parts and solving them: Tracking of the bat pose and localization of already detected landmarks; data association to single out measurements originating from the detected landmarks; and detection of new landmarks from the remaining measurements. By combining solutions to these three subtasks, a computationally cheap, precise algorithm for feature-based indoor mapping is obtained. Tests using real life data were performed to confirm the feasibility of the approach.