What the Robots See

Robots today are limited by their vision systems. Their vision systems lack the ability to accurately judge depth and determine the distance to an object. It is possible to load a robot with countless sensors to aid navigation through a field of objects. However, not all robots can take the additional sensors because of design space or cost issues. An alternative to extra sensors is to use a distance-vision algorithm that requires just one video camera. The algorithm uses specific vision cues to determine depth. Such cues include variations in texture, edges and haze. Lines that converge along their edges indicate increasing distances while hazy edges are objects that are farther away than objects with clear edges. The vision systems using this algorithm work well for slow moving robots where the distances are short. However, visual depth perception has high uncertainty at large distances, so robots moving at high speeds do not have the same success. There is a possibility that further research and development will overcome these challenges in the near future.  Read the full article at Mechanical Engineering Magazine.

Find out about McClellan's automation equipment capabilities.