A critical function for autonomous vehicle navigation equipment is the ability to maintain safe vehicle-to-vehicle (V2V) spacing between two moving vehicles. While GPS is the preferred method, the GPS signal can be disrupted by the presence of tall buildings and other obstacles, and/or by multipath reflection errors. And while lidar and radar are highly effective, they are also expensive to implement. An alternative method that is both accurate and inexpensive has been demonstrated on an experimental platform created by scientists at Hunan University (Changsha, China) using a simple monocular camera and visible light communication (VLC) equipment. Not only does VLC provide vehicle-to-vehicle positioning capability, but it also provides illumination and communication functions as well.
Immune to electromagnetic interference and consuming little power, the VLC equipment consists of a CPU and a monocular camera (an iPhone) mounted on the “estimating” vehicle that takes pictures of two LEDs spaced a fixed distance apart on the taillights of the “target” vehicle. Image processing, the pinhole model, and a Kalman filter algorithm is applied at the CPU to calculate the taillight positions relative to the estimating vehicle and continuously compute the V2V spacing parameters. For vehicles moving at speeds up to 22 m/s, the positioning accuracy can be maintained with centimeter-level accuracy. Reference: J. He et al., Opt. Express, 28, 4, 4433–4443 (2020).