This dissertation presents a hierarchical, hybrid point of view to the control of large-scale systems. The analysis is based on a new hybrid dynamical system formulation that allows for the modelling of large scale systems in a modular fashion. Three problems are addressed: controller design, closed loop performance verification and the extension of system autonomy. A control scheme based on semi-autonomous agent operation is first proposed. An algorithm, using ideas from game theory, is presented to produce continuous controllers for hierarchical, hybrid designs. A discussion of design issues involving controller autonomy is presented. The emphasis is on hybrid effects. The hierarchical and hybrid issues involved in the design of a fault tolerant control scheme are investigated. The study concludes with a case study on the application of the proposed techniques to the control of an automated highway system (AHS).

A simple model for a string of vehicles is constructed. The model explicitly accounts for the possibility of repeated collisions between the vehicles in the string. Based on the model a notion of safety is formulated for the string. Necessary and sufficient conditions are presented that specify when a string of vehicles is safe while performing a simple emergency deceleration maneuver where all vehicles start decelerating at a fixed rate after some delay. The conditions are interpreted in terms of their implications for the safety of platoons of vehicles.

We investigate the effect of a number of design alternatives on the safety and capacity of an Automated Highway System. Our methodology makes use of two computational tools, designed to highlight the fundamental limitations of the vehicle dynamics, sensing and control strategies and inter-vehicle communication. The first tool produces the minimum spacing necessary for two vehicles not to collide, as a function of their state and capabilities. The second tool investigates the multiple collisions that may occur in a string of vehicles if the spacing requirements of the first tool are violated. The example we have in mind is the emergency deceleration of a platoon of vehicles. We use the tools to establish limits on the safety and throughput that can be expected if different automated highway concepts are implemented. Our results indicate that on an Automated Highway System that supports platooning, severe intra-platoon collisions may occur under emergency situations. We analyze the effect of the collisions on both safety and throughput and investigate the sensitivity of these quantities to a number of design parameters. We identify as the most important parameters inter-vehicle coordination, on-line braking capability estimation and platoon organization in terms of braking capability.

Automation of highways and in particular platooning of vehicles raises a number of control issues. A hierarchical structure is used to address these issues. The work presented here is an attempt at constructing a consistent interface between discrete event and continuous time controllers. The design proposed is a finite state machine that communicates with the discrete controllers by issuing commands that get translated to "jerk" input for the vehicle engine. The operation of the proposed design is tested using COSPAN. By virtue of the fact that the interface touches on both the discrete and continuous worlds, the design might provide insight to interesting problems related to the hybrid nature of the system.

In this paper, a case study of the difficulties encountered in the design of hierarchical, hybrid control systems is presented. The Intelligent Vehicle Highway System (IVHS) architecture proposed for vehicle platooning is used as an example. The authors conclude that the conventional tools currently is use for the design and verification of control systems may be inadequate for the design of hierarchical controllers for hybrid systems. The analysis, while indicating certain shortcomings of the current IVHS design, also contains proposed solutions to these problems.

In this research, the authors focus on the problem of designing hierarchical, hybrid controllers for large scale systems. An overview of the proposed design process is first given. A modeling formalism is then introduced which allows for the modeling of multi-agent systems with hybrid dynamics in a modular fashion. An algorithm that makes use of ideas from game theory to capture the interaction between the agents, derive continuous controllers and quantify the discrete coordination needed to achieve certain performance objectives is presented. The algorithm is then applied to the problem of automated vehicle following. The study concludes with a summary of the proposed scheme and highlights of the major points.

In this paper, the authors propose a hierarchical control architecture for dealing with faults an adverse environmental conditions on an Automated Highway System (AHS). The design builds on a previously designed control architecture that works under normal conditions of operation. The faults that are considered in the design are classified according to capabilities remaining on the vehicle or roadside after the fault has occurred. Information about these capabilities is used by the supervisors in each of the layers to select appropriate control strategies. The extended control strategies which are needed by the supervisors of each layer of the hierarchy are outlined, and in certain cases, examples are given of their detailed operation.