A Simple And Modular Approach To Path Planning For Tractor-Trailer Robots Based On Modification Of Pre-Existing Trajectories

Abstract

Tractor-trailer Mobile Robot systems consist of a nonholonomic mobile robot regarded as a tractor and several passive trailers linked to the tractor via a hinge. As these systems are in many cases more economical compared to multi mobile robot systems, they are nowadays used for the transport of various objects in the field of logistics. Hence an exact and efficient path planning algorithm is required. Due to the highly nonlinear characteristics of such a system, path planning is always a challenging problem. Many path planning algorithms have already been proposed for mobile robots. While some of these approaches can solve the path planning problem for robots with multiple trailers, the solutions are usually complex, limited to a specific hardware configuration, or very computationally expensive. In this work, we present an algorithm that, although still computationally expensive in its current form, provides a very simple, extensible, and flexible solution for path planning of tractor-trailer robots. Based on a conventional global planner for mobile robots (like A* or Dijkstra), our algorithm adjusts the global path according to the dimension of the carried object so that the whole system traverses a collision-free path. We take advantage of the fact that the global planner has already planned a collision-free path for the mobile robot (tractor), which the trailer follows almost exactly on straight paths. Accordingly, collisions occur predominantly in or shortly after curves. We designed our algorithm to detect these particular curves and adjust the curve radius so that no collision occurs. This way, we do not need to re-plan the whole trajectory. Also, we can upgrade almost any planning algorithm for mobile robots to work with tractor-trailer systems. We validate our method based on a series of simulations. First real-world experiments are also very promising.