A global-local approach for optimal trajectory generation
MetadataShow full item record
Optimal trajectory generation is an important task for mobile robot navigation in an unknown environment, like in Mars. This thesis proposes Global-Local Orthogonal Mapping algorithm ( GLOMAP ) into rough terrain cases and quadrotor aerial case. GLOMAP algorithm approaches the global domain with multiple local domains in each time interval. This method replaces the nonlinear global domain problem into solving local domain problems. Alternatively, the preliminary local approximation can be chosen arbitrarily to take advantage of prior knowledge of a particular problem. Potential function theory is used to represent the allowable slope boundary in the terrain attempts. In a rough terrain case, four three-dimensional terrain environments are posted for ground vehicle to solve the fuel optimal trajectory generation problem. The optimal trajectories and fuel cost values are compared with SNOPT and fmincon : two different nonlinear optimizers. In a 3D space case, the time optimal shortest trajectory is observed. The optimal time value and optimal trajectory are compared with the paper reference results. To visualize the ground vehicle performance in 3D rough terrain, Player/Gazebo simulation environment is utilized to display the ground vehicle motion movement in rough terrain. Although the GLOMAP algorithm proposed here is designed for optimal trajectory, it also can be applied to other problems with little or no variation.