Stereo-vision framework for autonomous vehicle guidance and collision avoidance

摘要: During a pre-programmed course to particular destination, an autonomous vehicle may potentially encounter environments that are unknown at the time of operation. Some regions contain objects or vehicles were not anticipated during mission-planning phase. Often user-intervention is possible desirable under these circumstances. Thus it required for onboard navigation system automatically make short-term adjustments flight plan and apply necessary corrections. A suitable path visually navigated through environment reliably avoid obstacles without significant deviations from original course. This paper describes general low-cost stereo-vision sensor framework, passively estimating range-map between forward-looking its environment. Typical be either unmanned ground airborne vehicles. The image relative distance observed contains information could used compute navigable plan, also visual geometric detail about other processes future missions. Aspects relating flow framework discussed, along with issues such as robustness, implementation advantages disadvantages framework. An outline physical structure presented overview algorithms applications given.