Near Optimal Trajectory Generation for Omnidirectional Vehicles by Constrained Dynamic Inversion
Additional Document Info
This paper presents a computationally cheap algorithm for near-optimal trajectory generation in presence of control constraints. We consider the problem of moving an omnidirectional vehicle from a specified initial state to a desired final state while maintaining a good trade-off between the maneuver time and the final state-error. A parameterized spline that satisfies the boundary conditions of the problem is used as an reference trajectory. The reference trajectory is parameterized by total maneuver time which also serves as an optimization parameter. Dynamic inversion is used to calculate the control required for tracking the reference trajectory. However, perfect tracking is not possible because of actuator saturation and noise. Therefore a tracker is implemented to set reasonable error-dynamics between the commanded and achieved trajectories. An optimization problem is posed to achieve the desired trade-off between the total time and final state-error. The performance of the algorithm is evaluated by comparing the generated trajectories with benchmark solutions obtained from a Chebyshev based pseudo-spectral optimization method.
name of conference
AIAA Guidance, Navigation and Control Conference and Exhibit