A Novel Predictor Based Framework to Improve Mobility of High Speed Teleoperated Unmanned Ground Vehicles

[摘要] Teleoperated Unmanned Ground Vehicles (UGVs) have been widely used in applications when driver safety, mission eciency or mission cost is a major concern. One major challenge with teleoperating a UGV is that communication delays can significantly affect the mobility performance of the vehicle and make teleoperated driving tasks very challenging especially at high speeds. In this dissertation, a predictor based framework with predictors in a new form and a blended architecture are developed to compensate effects of delays through signal prediction, thereby improving vehicle mobility performance. The novelty of the framework is that minimal information about the governing equations of the system is required to compensate delays and, thus, the prediction is robust to modeling errors.This dissertation first investigates a model-free solution and develops a predictor that does not require information about the vehicle dynamics or human operators;; motion for prediction. Compared to the existing model-free methods, neither assumptions about the particular way the vehicle moves, nor knowledge about the noise characteristics that drive the existing predictive filters are needed. Its stability and performance are studied and a predictor design procedure is presented.Secondly, a blended architecture is developed to blend the outputs of the model-free predictor with those of a steering feedforward loop that relies on minimal information about vehicle lateral response. Better prediction accuracy is observed based on open-loop virtual testing with the blended architecture compared to using either the model-free predictors or the model-based feedforward loop alone. The mobility performance of teleoperated vehicles with delays and the predictor based framework are evaluated in this dissertation with human-in-the-loop experiments using both simulated and physical vehicles in teleoperation mode. Predictor based framework is shown to provide a statistically significant improvement in vehicle mobility and drivability in the experiments performed.

[发布机构] University of Michigan

[效力级别] delay compensation

[关键词] Teleoperated UGVs;delay compensation;predictor based framework;vehicle mobility;Mechanical Engineering;Engineering;Mechanical Engineering