A Method for Modeling and Prediction of Ground Vehicle Dynamics and Stability in Autonomous Systems

A future limitation of autonomous ground vehicle technology is the inability of current algorithmic techniques to successfully predict the allowable dynamic operating ranges of unmanned ground vehicles. A further difficulty presented by real vehicles is that the pay-loads may and probably will change with unpredictably time as will the terrain on which it is expected to operate. To address this limitation, a methodology has been developed to generate real-time estimations of a vehicle’s Instantaneous Maneuvering Manifold. This approach uses force-moment method techniques to create an adaptive, parameterized ve-hicle model. A technique is developed for estimation of vehicle load state using internal sensors combined with low-magnitude maneuvers. An unscented Kalman filter based estimator is then used to estimate tire forces for use in determining the ground/tire co-efficient of friction. Probabilistic techniques are then combined with a combined-slip pneumatic trail based estimator to estimate the coefficient of friction in real-time. This data is then combined to map out the instantaneous maneuvering manifold while applying techniques to account for dynamic rollover and stability limitations. The algorithms ar