Robots extract highly recognizable information from real-time images via vehicle-mounted camera, and match these information with the off-line map, so as to know their location in the map and which direction to go in the next step.

Visual information is the most extensive among all sensor information. We can extract aplenty information from the environment to build the environment map with multiple dimensions and modals, so as to realize positioning and navigation algorithm with stronger stability and better adaptability.

Diversified navigation solution for different scenarios； The cost of visual sensors are much lower than the same level laser sensors； Construction-free, or just some QR codes and slight adjustments； Quick implementation, low maintenance cost； High modularity, easy to retrofit from existing vehicles