On-Line Path Generation and Tracking for High-Speed Wheeled Autonomous Vehicles
Abstract
In this work we proposed two semi-analytic methods to generate minimum and near-minimum time velocity profiles for a vehicle along a specified path. Initially we adopt a point mass parameterization of the vehicle with specified acceleration limits. In generating the optimal velocity profile, several undesirable cases, where loss of controllability occurs, and which have been neglected in the literature, are dealt with in this work. A receding horizon implementation is also proposed for the on-line implementation of the velocity optimizer. Robustness of the receding horizon algorithm is guaranteed by the use of an adaptive scheme that determines the planning and execution horizons. Application to a Formula 1 (F1) circuit with a comparison between the infinite and finite receding horizon schemes provides a validation of the proposed methodology. We also provide extensions from the point mass to a half-car model to recover the missing attitude (yaw) information. The acceleration limits (GG-diagram) of the half-car model is determined by the available tire friction forces in the front and rear axles. We present three extensions of the point mass methodology to the half-car model. In the first extension we directly implement the optimal control strategy of the point mass case to the half-car model. In the second extension the optimal control strategy of the point mass case is interrupted by a stabilizing control logic when the vehicle slip angle increases beyond a certain value and the yaw dynamics tend to instability. Finally, in the third approach we enforce the additional constraint that the vehicle tracks the path with zero slip angle and determine the acceptable acceleration limits subject to the new constraint.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 17, 2006
- Accession Number
- ADA447359
Entities
People
- Panagiotis Tsiotras
Organizations
- Georgia Tech