Range imaging is the measurement of the three-dimensional shape of an object as an array of 3-D distance values. While range imaging is a critical robotic sensing technology, existing methods are very slow; the best current devices still produce only a few tens of frames per second. We propose to develop an ultra fast range imaging system consisting of a scanning laser and a sensor array of photosensitive cells. The scanning laser sweeps a scene continuously with a moving light stripe, and each sensor cell of the array measures the distance by recording the time at which the incident light intensity peaks during each sweep. The novelty of this system is a VLSI-based computational sensor which exploits the inherent parallelism of this techniques by integrating photoreceptors, analog circuitry, and digital logic on a single CMOS die. The resulting system can produce 1000 range images per second, two orders of magnitude faster than current systems. Also, the system will be simple and compact since it does not require a separate computer for processing signals. Phase I of this project will develop a prototype design that will lead to commercial production of a highly advanced, high-speed range imager. ANTICIPATED
Benefits: An ultra fast, 3D range imager will revolutionize the machine vision market, traditionally dominated by 2D techniques. Fast 3D vision and measurement systems will have endless applications, ranging from military and manufacturing to robotics, medicine, and textiles. Examples include: Whole shape measurement of a vehicle for radar reflectance simulation; design and inspection of manufactured parts; tele-existence via 3D images; and 3D imaging of the human body for reconstructive surgery, design of protective equipment, and the fashion industry.
