Geometric approach to the solution of the Dubins car problem in the formation of program trajectories

Abstract

The paper considers an approach to the formation of control program trajectories of moving objects (UAVs, ships) as a solution to the optimal problem in terms of Dubins path search. Instead of directly solving the Pontryagin’s maximum principle, it is proposed to use a simple analysis of possible control strategies in order to determine among them the optimal one in terms of time spent on a trajectory. The problem of finding the shortest trajectory of movement of an object from one point to another is solved, and for both points their coordinates and heading angles at these points are given, as well as three absolute values of the circulation radii corresponding to the given control signals on each of the three sections of the trajectory. The problem of finding the Dubins curves is reduced to determining the parameters of two intermediate points at which the control changes. All possible directions of control change options are considered, taking into account the existing constraints, also the lengths of the corresponding motion trajectories are calculated, and the optimal one is selected. The problem of constructing a trajectory is solved as well, which ensures a smooth conjugation of two linear fragments of trajectories and passes through the point of their intersection. The solution of the optimal trajectory problem using the Dubins car gives a single trajectory. In contrast to this, the proposed method considers several trajectories admissible by the constraints, from which the optimal one is selected by exhaustive search. The presence of several feasible strategies gives advantages for each specific situation of choosing a trajectory depending on the environment. Instead of directly solving the Pontryagin’s maximum principle and constructing a three-dimensional optimal trajectory, the authors used a simple analysis of possible control strategies in order to determine among them the optimal one in terms of elapsed time. The approach was motivated by the limited number of possible control strategies for Dubins paths, as well as the simplicity of analytical calculations for each of them, which allows performing these calculations in real time. The high speed of calculations for the problem of determining the optimal trajectory is due to the fact that the proposed method does not require complex calculations to solve the problem of nonlinear optimization, which follows from the Pontryagin’s principle.

Keywords

• dubins car
• trajectory
• algorithms
• optimal control
• conjugation curves