On the way to autonomous driving, advanced driver assistant systems (ADAS) and automated driving function (ADF) represent essential parts of the puzzle. Their objective is to improve safety and driving comfort by assisting the human driver. Automated control systems regulate parts of driving functions, advisory and warning systems preserve the driver from potential accidents or relieve him from tedious tasks during monotonous driving events. Some of these functions like adaptive cruise control (ACC) systems taking over longitudinal vehicle control have already been successfully implemented in standard production vehicles. Nevertheless, ADAS still provide a wide area of research and development for further improvements. Recent investigations have demonstrated that ACC contains a lot of potential for increasing driving comfort and economy by applying anticipatory control actions. Moreover, modern communications technologies allowing information exchange between vehicles offer the investigation of advanced control design concept which provide a great positive effect on traffic flow characteristics like traffic flow stability.
Against this background, this thesis proposes a design and comparison study on two different control concepts applied to ACC and lane change assisted ACC (LCACC). Both strategies rely on idealized connected environment assumptions where vehicles and infrastructure transmit information about their actual states via modern communication technologies. The controlled vehicle (denoted as the ego-vehicle) is able to receive and utilize this information within the control design process. Mixed H2/H∞ control concepts are proposed primarily targeting traffic flow or string stability and its robustness. Thereby, a novel concept which is entitled platoon string stability is proposed: Human driving behavior tends to show unstable characteristics during car following situations from a traffic flow perspective. Considering a vehicle platoon where human drivers are steering the individual vehicles the introduced ego-vehicle controller counteracts this unstable behavior and accomplishes string stability of the platoon, i.e. the transfer behavior from the foremost vehicle to the ego-vehicle remains string stable although individual vehicle in between present unstable characteristics. As an alternative ACC strategy, predictive control methods are investigated targeting driving comfort and economy of the ego-vehicle. Thereby, model predictive control (MPC) techniques utilizing a prediction model to estimate the future driving behavior of leading vehicles are suggested.
The prediction allows to accomplish anticipatory driving actions of the ego-vehicle and shows to increase driving comfort and economy significantly compared to existing ACC control approaches or human driving behavior. Both ACC control concepts are then extended to cope with multi-lane traffic situations by providing active lane change control besides the longitudinal vehicle control, denoted as lane change assisted adaptive cruise control (LC-ACC). A minimum entropy concept is applied for the lane decision process within the mixed H2/H∞ control strategy. This approach allows to address traffic flow stability within multi-lane traffic situations and balances local differences in traffic flow characteristics between the individual lanes. The predictive control strategy focuses on ego-vehicle objectives like desired velocity tracking and driving comfort similar to the ACC case. An extended evaluation and comparison study utilizing an advanced co-simulation framework which combines vehicle and traffic simulation tools reveals benefits and drawbacks of the different ACC and LC-ACC concepts. On the one hand, the predictive strategies present advantages concerning driving comfort and economy compared to the mixed H2/H∞ control approaches. However, they do not consider the effect of their control actions on surrounding traffic participants which may have severe impact of string unstable traffic flow, e.g. due to egoistic lane change maneuvers. In contrast to that, the mixed H2/H∞ strategies achieve not only string stable ego-vehicle behavior but improve the overall traffic flow stability within a certain area around the ego-vehicle which represents a clear benefit over the predictive control approaches.