Evolutionary trajectory learning for autonomous robots by means of geometric approximations of polygonal obstacles

Addresses the issue of flexible geometric approximations of polygonal obstacles for intelligent autonomous robot (IAR) navigation which is an extension of our previous work (1998). The trajectory learning problem for IAR navigation is formulated as a constrained discrete-time-optimal control problem where the polygonal obstacles are the constraints. From the visibility and sensor modeling concepts, polygonal obstacles within the environment are approximated as either by circles or by ellipses depending on the shape and size of them. Furthermore, some practical issues are identified and resolved through these type of approximations. The effectiveness of these methods is illustrated by some simulations of the robot within a heavily obstacled environment