The imagery from three cameras is fused in order to collect a hemispherical view of the environment around train tracks. An embedded FPGA ensures that the three heads acquire the frames simultaneously, while also ensuring that each head is properly exposed to within a given tolerance to the other heads — think of the sun illuminating from one side more than the other. With the help of a precision IMU and GPS receiver, the frames are timestamped with a 1 microsecond resolution, allowing accurate geolocation even on a fast moving platform such as a train.
