Integrated Visual and Inertial Sensor Information for Onboard Navigation (I-VISION)

Given the vulnerability of the GPS system to intentional or hostile denial of service, it is imperative that the U.S. military develop technologies that provide robust navigation solutions that do not rely on GPS data. To achieve this capability, Aurora and MIT have teamed to develop the Integrated Visual and Inertial Sensor Information Onboard Navigation (I-VISION) system, which will couple optic flow-based ego-motion estimation with inertial navigation systems to achieve precision navigation without GPS.. Current-generation inertial measurement units (IMUs) provide very accurate angular rate and linear acceleration information, which can be integrated to estimate a platform’s position, velocity, and orientation over time. Unfortunately, due to sensor biases and scale factor errors, the accuracy of the estimation degrades with time. The coupling of visual flow and IMU-based ego-motion estimation enables the IMU to aid in resolving vision-based errors and ambiguities, and visual flow estimates can be used to reduce IMU sensor biases and scale factor errors. A multiple-camera system will be considered in order to better address the ambiguity that is introduced in the estimation of optical flow-based ego-motion. Both tightly coupled and ultra-tightly coupled implementations will be considered.