Optimisation based motion planning for non-holonomic vehicles in narrow environments

Más información sobre el libro

This work is concerned with path planning and motion control for non-holonomic vehicles. In particular, path planning for a car for parking scenarios and for a truck with attached trailer at a loading bay is treated. In this context, challenges arise from the close vicinity to obstacles and the non-holonomic characteristics of both vehicles, i. e. a pure sidewards movement and turning without moving are not possible. To achieve a real-time capable solution, global planning with an in advance unknown, arbitrary number of driving direction changes is not possible. Thus, a local approach is presented which basic idea is to mimic the behaviour of the human driver. This yields habitual paths which are applicable for practical implementation. A global guidance scheme ensures high convergence rates in even complex scenarios like on parking decks. To obtain a feasible trajectory, an optimal time parametrisation respecting the limitations of the vehicle is presented. To suppress disturbances and account for model uncertainties, a Lyapunov-based feedback control scheme is introduced. This control scheme does not apply additive inputs which minimises the risk of collision with obstacles.