A Novel Omnidirectional 3D Camera for Battle Field Modeling and Reconnaissance Vehicles

To the best of our knowledge, there is no existing battle field modeling or reconnaissance technology that can provide both the real-time three dimensional (3D) range information and the omni-directional viewing capability at any cost. The primary objective of this project is to fill this gap, and to develop a high speed and low-cost 3D image sensor with an omni viewing angle (i.e., 360 degree) for the battle field modeling and reconnaissance tasks in digital battle field management applications. The omnidirectional vision technique offers an extremely wider angle field of view. This sensing capability provides the autonomous vehicles with advantages of no dead viewing angle area, a faster response time and a high speed detection of targets and/or possible intrusion in all directions, lower installation and operating costs, and suitable for a variety of practical battle field environments. The proposed system, called OMNI 3D Camera, is able to provide full frame 3D images at a rate of 30 frames per second or higher, a capability that no other commercially available 3D sensor can achieve at any cost. The omnidirectional 3D range image can be obtained in a single snapshot with no scanning operation or mechanical moving part. This novel 3D sensoring approach eliminates the computationally expensive process for integrating multiple images captured in different viewing angles, therefore the throughput of the usable 3D information is further enhanced. No scanning nor mechanical moving part is needed. Not only the mechanical design becomes very simple and reliable, but also the no-scan feature overcomes the inherent error source in scanning laser based systems due to the high speed motion of object during scanning. Most components in this system are Commercial Off The Shelf (COTS) products, which leads to low installation/operation cost and short development time. Our Omni 3D camera can also provide normal 2D color image of the scene around the autonomous vehicle using the same CCD camera. The registered 2D and 3D images facilitate the an easy implementation of sensor fusion and object recognition algorithms. Depending on specific application scenario, the color spectrum of the light projector can be s