Developments in Estimation and Control for Cloud-Enabled Automotive Vehicles.

Abstract

Cloud computing is revolutionizing access to distributed information and computing resources that can facilitate future data and computation intensive vehicular control functions and improve vehicle driving comfort and safety. This dissertation investigates several potential Vehicle-to-Cloud-to-Vehicle (V2C2V) applications that can enhance vehicle control and enable additional functionalities by integrating onboard and cloud resources.

Firstly, this thesis demonstrates that onboard vehicle sensors can be used to sense road profiles and detect anomalies. This information can be shared with other vehicles and transportation authorities within a V2C2V framework. The response of hitting a pothole is characterized by a multi-phase dynamic model which is validated by comparing simulation results with a higher-fidelity commercial modeling package. A novel framework of simultaneous road profile estimation and anomaly detection is developed by combining a jump diffusion process (JDP)-based estimator and a multi-input observer. The performance of this scheme is evaluated in an experimental vehicle. In addition, a new clustering algorithm is developed to compress anomaly information by processing anomaly report streams.

Secondly, a cloud-aided semi-active suspension control problem is studied demonstrating for the first time that road profile information and noise statistics from the cloud can be used to enhance suspension control. The problem of selecting an optimal damping mode from a finite set of damping modes is considered and the best mode is selected based on performance prediction on the cloud.

Finally, a cloud-aided multi-metric route planner is investigated in which safety and comfort metrics augment traditional planning metrics such as time, distance, and fuel economy. The safety metric is developed by processing a comprehensive road and crash database while the comfort metric integrates road roughness and anomalies. These metrics and a planning algorithm can be implemented on the cloud to realize the multi-metric route planning. Real-world case studies are presented. The main contribution of this part of the dissertation is in demonstrating the feasibility and benefits of enhancing the existing route planning algorithms with safety and comfort metrics.