Vehicle-oriented control theoretic research – inaugural lecture by Prof. Péter Gáspár
The control of the key elements of transport (e.g., of vehicles and of vehicle platoons) is normally carried out in a hierarchical manner with the various elements of control targeting different layers of the hierarchy.
In case of a single vehicle, the control methods are relied on and used in the design of its components and functions, while in case of a cooperative vehicle platoon, the task is to design the cooperative control between, over and amongst the vehicles. The vehicle-oriented R&D work involves the application of electronic and automation functions, relies on the persistent improvement of the sensor and actuator characteristics and features, makes use of the high-end information and telecommunication techniques, and last but not least, applies the scientific methods of system and control theory that enable the researchers and developers to explore, understand the particular vehicular system requirements and to design high-quality useable vehicular systems.
In the research carried out in this field, the model-based and robust control design principles play an important role. These principles include the Newtonian and Lagrangian characterisation of motion dynamics, the control-oriented modelling that takes the quality specifications and the model uncertainties into consideration, and the control design methodologies ‒ based on these principles ‒ that target various model classes.
In our research, we applied linear time-invariant (LTI) models that are capable of tackling uncertainties, and state- and parameter-dependent models which belong to various linear parameter-varying (LPV) model classes and are capable of tackling non-linear features and system elements. In case of most vehicle-oriented R&D and engineering tasks, the simultaneous optimisation in regards of the diverse quality specifications is out of question, therefore finding the middle ground between these requirements and specifications is of great importance. Such a compromise is normally achieved in the modelling and design via application of appropriately chosen weighting functions. For these reasons, we had to rely on robust control design methods that guarantee the fulfilment of certain quality requirements, and at the same time take the model and parameter uncertainties and the various disturbances into consideration.
In the following, we identify and present four main streams of the current vehicle-oriented control theoretic research. These are the traditional control of a single vehicle, the integrated control of a single vehicle, the cooperative control that involves a group of vehicles (e.g., a platoon), and the autonomous vehicle control (presently mostly of a single vehicle only).
The traditional control tasks ‒ in regards of a single vehicle ‒ aim for improving its components and functions, as well as their effectivity. The challenges posed by these tasks often lead to new design methods and algorithms.
The integrated control aims at harmonising the reliance on system resources (e.g., sensors, actuators, local facilities of control) within the vehicular system and at handling the interactions between the specific control mechanisms, which were by default designed independently of one another. In case of integrated control designs, the prioritisation of the tasks depending on vehicular states must be implemented and adhered to, or in other words, the hierarchy between the various actuators must be maintained. Methods that can adapt to altered circumstances and environment can be realized using the reconfiguring control methodology. The consequences of the fall out of a component from service can be mitigated by using design approaches implementing fault-tolerant control.
The cooperative control aims to harmonise the separate control mechanisms used in conjunction with vehicles ‒ within a group of vehicles ‒ in order to achieve common group-level goals, such as the group-level reduction of harmful emissions, and of fuel-consumption. Cooperative control solutions include ‒ among other quite diverse approaches ‒ control mechanism that makes use of a parametrised driver model, and one that computes optimal speed profile for a route considering the geographical and meteorological environment of the road and other road and traffic features (e.g., junctions, traffic lights and speed limit signs, traffic levels at different times).
Probably the most important and most challenging problem within the vehicle-oriented control research field is the autonomous vehicle control. In the evaluation of the vehicle-environment interaction, the environment detection methods and the methods providing and increasing situation awareness play a very important role. In case of autonomous vehicular control design, the prescribed trajectory ‒ or the trajectory modified due to environmental changes ‒ must be followed normally. Several examples from our own research can be given in this respect.
The video recording of the Inaugural Lecture at HAS can be viewed here. The lecture was delivered in Hungarian, unfortunately no English subtitles are available.