Modelling Human-Driver Behaviour Using a Biofidelic Approach

Martínez García, Miguel (2018) Modelling Human-Driver Behaviour Using a Biofidelic Approach. PhD thesis, University of Lincoln.

Full content URL:

Modelling Human-Driver Behaviour Using a Biofidelic Approach
[img] PDF
Martínez Miguel - Engineering - December 2018.pdf - Whole Document

Item Type:Thesis (PhD)
Item Status:Live Archive


This dissertation is concerned with the subject of modelling human steering control of ground vehicles. Special care has been taken with respect to designing a model that is biofidelic, i.e., a model that operates according to the principles of human control. With this aim, first classical human control theory has been revisited, both from a literature review and an experimental perspective; data have been recorded from test subjects in compensatory and pursuit tracking tasks. The tracking experiments are the first ever to be performed with fractional order plants, which are plants suitable to represent system memory. From the data, an extension of the Crossover model by McRuer’s is designed, to include the control of such category of plants. The proposed model is referred to as the Fractional Crossover Model. This is followed by a study on modelling memory in human-machine systems from a classical control theory viewpoint. These results broaden the existing array of manual control modelling techniques and can be employed in a modular manner, combined with current models. More significantly – and still with respect to the domain of generic human control and human-machine systems – a new approach for modelling the human-operator is proposed. This approach consists in treating the problem from a statistical viewpoint. With this methodology a novel human control model based on multiplicative dynamics is presented. The model, which was inspired on actual results in neuroscience, is validated with the tracking data obtained from test subjects and by comparing it to classical models in the literature. Hence the model is useful to analyse human performance or to reproduce human control in simulation, field tests or in the video game industry. With respect to steering control modelling, which is the main topic of this dissertation, additional experiments with test subjects were conducted in a simple vehicle simulator – with hardware and software specifically developed during this research program to test multiple hypotheses. The data were analysed with the intent of identifying which optical variables drivers employ while controlling a vehicle on public roads; it is seen that the splay angles– which are the projections of the road lines on the retina – are likely candidates for lane keeping at low speeds. This brings on a novel human-centred driver model first proposed here. This model includes multiplicative human control over the splay angles, and far-point error perception for lane keeping at higher speeds. The human-centred model has its domain of applicability in the intelligent transportation industry, in particular for the development of shared control systems and advanced driver-assistance systems for semi-autonomous ground vehicles. Additionally, the model can be employed in field testing of ground vehicles – for example, in vehicle durability tests. Furthermore, the topic of alternative steering devices for driving autonomous and semi-autonomous vehicles is investigated. This leads to another of the contributions in this dissertation. Here it is proposed that for such vehicles, and for the control of systems with a shared control perspective, anisometric steering wheel can be advantageous under certain schemes – tight rein or loose rein modes according to the H-metaphor. This is supported by additional data collected in the driving simulation experiments. Resulting from this, fractional order transfer functions are employed to increment steering stability and control accuracy with the isometric device. This prototypical steering system is applicable for the control of ground vehicles with the so-called by-wire controls, which are already incorporated in some commercially available vehicles.

Divisions:College of Science > School of Engineering
ID Code:37686
Deposited On:07 Oct 2019 11:40

Repository Staff Only: item control page