Motion Planning for Multiple Mobile Robots Using Time-Scaling

Abstract

Recent developments in robotics have raised the interest in mobile robots due to their wide range of applications requiring high level of autonomy of individual mobile agents. This increased level of autonomy needs more sophisticated path-planning and motion control methods, which have been studied extensively in the recent years This paper focuses on a typical industrial application in which several mobile robots share the same workspace. These units may carry materials or be capable of carrying out various repair jobs at some locations, etc. Due to the tight schedules to achieve high productivity in such applications, it is imperative to synchronize the motion of the units so that they arrive at the required location on time without collision with obstacles or with other units. Coordinated planning of such multi-robot systems is addressed in the literature based on some search methodology [7] or using neural network based techniques Since a high number of units can be involved, it may not be possible to design collision free trajectories for all robots, i.e. avoiding each other and all static obstacles cannot be guaranteed only by the path geometry. Instead of constructing trajectories with complex geometries so that they have only a few or no intersections at all, one can design simple paths which disregard moving obstacles (i.e. other robots) and take only the static obstacles into account, so the path geometries are supposed to be fixed for each robot. Such motion planning algorithms are readily available in the literature This initial path geometry designed and assigned to each unit ensures the avoidance of static obstacles. The avoidance of the moving obstacles can be achieved by choosing a suitable velocity profile along the fixed-geometry trajectories. In order to obtain the proper velocity input, we assign first a default velocity profile which the unit would use without the dynamic obstacles. Such a profile can be designed to be optimal in some sense (minimal time or cost) and may satisfy additional constraints, if necessary The default velocity profile consists of intervals with constant or zero acceleration. Acceleration parameters change at the boundaries of these intervals. This default velocity profile is then modified such that a unit decelerates or accelerates its motion to let other units pass through potential collision areas (identified from the path geometries). We suppose that a priority level is assigned to each robot so that a unit with a given priority regards all units with higher priority level as dynamic obstacles. The ranking of the units is www.intechopen.com