Motion planning and control for articulated logistic vehicles such as tugger trains is a challenging problem in service robotics. The case of tugger trains presents particular difficulties due to the kinematic complexity of these multiarticulated vehicles. Sampling-based motion planners offer a motion planning solution that can take into account the kinematics and dynamics of the vehicle. However, their planning times scale poorly for high dimensional systems, such as these articulated vehicles moving in a big map. To improve the efficiency of the sampling-based motion planners, some approaches combine these methods with discrete search techniques. The goal is to direct the sampling phase with heuristics provided by a faster, precociously ran, discrete search planner. However, sometimes these heuristics can mislead the search towards unfeasible solutions, because the discrete search planners do not take into account the kinematic and dynamic restrictions of the vehicle. In this paper we present a solution adapted for articulated logistic vehicles that uses a kinodynamic discrete planning to bias the sampling-based algorithm. The whole system has been applied in two different towing tractors (a tricycle and a quadricycle) with two different trailers (simple trailer and synchronized shaft trailer).
Discrete Search Leading Continuous Exploration for Kinodynamic Motion Planning