Automatic reparking of the robot trailer along suboptimal paths

The work investigates the experimental problem of reparking trailer for a wheeled robot with a trailer. The geometric model with kinematic constraints leads to the sub-Riemannian problem. We solve this problem via nilpotent approximation. The corresponding solution is close to optimal and locally minimizes the total kinetic energy of the driving wheels. A full-scale model is designed in a way to avoid phase constraints usually appearing in trailer systems. We perform 64 different experiments with reparking trailer and obtain the satisfactory accuracy: for one maneuver we repark the trailer with maximum angle error equal to 4 degrees.