Robot navigation from a Gibsonian viewpoint

摘要

Gibson's insights into how optic flow can be used by animals to guide their actions have been formalized to some extent in Warren's laws of control. We noted that these laws are applicable to any moving agent, thus this approach could be very successful in the domain of behavior-based robotics. To demonstrate this we devised control laws for the obstacle avoidance problem in mobile robotics. Two laws of control relevant to the problem were proposed and tested on an actual robot in an unmodified office environment. The success and occasional, yet understandable, failures of these two laws show that the Gibsonian approach to visually guided navigation is very promising.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>