Relative and Absolute Localization Techniques
Southwest Research Institute (SwRI) is developing a variety of relatively low-cost localization technologies and related mapping approaches for use with ground vehicles. SwRI has developed both hardware systems and software algorithms to solve various aspects of the localization problem. A number of these systems have been used and demonstrated on full-scale military robotic vehicles for the USMC and U.S. Army.
SwRI makes use of a wide variety of technologies and methods that can provide for accurate, precise, reliable, and convenient localization in a variety of environments, from urban on-road, to austere off-road areas and in a wide range of weather and illumination conditions. In particular, these approaches address some of the key limitations of GPS, such as its susceptibility to denial and jamming by providing alternative global localization methods and by minimizing the drift of relative localization methods. The methods that have been developed, used, and evaluated by SwRI include traditional inertial navigation and odometry, visual odometry, and several monocular and stereo camera-based place recognition techniques.
SwRI developed integrated localization systems for the USMC and the Army that provide robust localization capabilities during significant GPS dropouts. The low-drift relative localization approach has allowed a driverless vehicle to drive a complex route autonomously for several kilometers without GPS. SwRI has also developed and tested a camera-based localization system that can provide day and night localization precision of less than 1 cm.