How does a racer drive around a track? Approaching a bend in the road, a driver needs to monitor the road, steer around curves, manage speed and plan a trajectory avoiding collisions with other cars - and all of this, fast and accurately. For robots this remains a challenge: despite progress in computer vision over the last decades, artificial vision systems remain far from human vision in performance, robustness and speed. As a consequence, current prototypes of self-driving cars rely on a wide variety of sensors to palliate the limitations of their visual perception. One crucial aspect that distinguishes human from artificial vision is our capacity to focus and shift our attention. This project will propose a new model of visual attention for a robot driver, and investigate how attention focusing can be learnt automatically by trying to improve the robot's driving. How and where we focus our attention when solving a task such as driving is studied by psychologists, and the numerous models of attention can be sorted in two categories: first, top-down models capture how world knowledge and expectations guide our attention when performing a specific task; second, bottom-up models characterise how properties of the visual signal make specific regions capture our attention, a property often referred to as saliency. Yet, from a robotics perspective, there remains a lack of a unified framework describing the interplay of bottom-up and top-down attention, especially for a dynamic, time-critical task such as driving. In the racing scenario described above, the driver must take quick and decisive action to steer around bends and avoid obstacles - efficient use of attention is therefore critical. This project will investigate the hypothesis that our attention mechanisms are learnt on a task specific basis, in a such a way as to provide our visual system optimal information for performing the task. We will investigate how state-of-the-art computer vision and machine learning approaches can be used to learn attention, perception and action jointly to allow a robot driver to compete with humans on a racing simulator, using visual perception only. A generic learning framework for task-specific attention will be developed that is applicable across a broad range of visual tasks, and bears the potential for reducing the gap with human performance by a critical reduction in current processing times.

In 2005, the Vehicle and Operator Services Agency (VOSA) introduced a computerised system for reporting MOT (roadworthiness) test results. Since that time, the results of approximately 35,000,000 MOT tests annually have been collected and stored in a Department for Transport (DfT) database. The DfT business plan , published 8 November 2010, promised to make available the "detailed VOSA MOT data" - and on 24 November, comprehensive data was released - consisting of the results of 150,000,000 MOT tests from 2005 to the spring of 2010. Some fields, such as vehicle registration plates and unique VTS (vehicle test station) identities have been withheld from the published data in order to preserve anonymity. However, what remains still contains a wealth of information that is not available in any other data set. In addition to the results of the MOT test itself (including detailed reasons for failure), the data include: - the vehicle odometer (mileage) reading - the vehicle manufacturer, type and engine capacity - the vehicle's year of first use - the top-level postal area (letters only from the postcode) of the VTS Our initial objective is to use the vehicle odometer readings - which are not available in any other (large scale) data set - combined with the data about vehicle type, to analyse how patterns of vehicle usage (and associated carbon footprint) have changed with time, disaggregated over different regions of the country. The project will therefore aim: - to develop software tools for the analysis of the MOT data; - to work with the DfT and VOSA on maximizing the use that can be made of the MOT data set whilst respecting issues such as data protection; - to scope the application of MOT odometer readings and the possibilities for triangulating with other data sets (such as vehicle emissions, new vehicle registrations and Census data); - to develop one (or two) small-scale demonstrations illustrating potential applications of our approach. The ultimate aim, going beyond the scoping study, is to create a publicly available tool that all those undertaking travel behavior change initiatives could use to assess the impacts of their work on car ownership, use and related carbon emissions, thereby dramatically reducing the need for every individual project to commission surveys or other forms of travel behavior measurement. Further research could also include specific analyses of: changes in car ownership and use that have occurred in the Sustainable Travel and Cycling Demonstration Towns; the nature of the distribution and diffusion of electric, hybrid and other alternative-technology vehicles; the location and concentration of 'dirty' vehicle use with implications for the targeting of climate change and air quality initiatives; and the relationship between car use and physical activity.

We propose a new phase of the successful Mathematics for Real-World Systems (MathSys) Centre for Doctoral Training that will address the call priority area "Mathematical and Computational Modelling". Advanced quantitative skills and applied mathematical modelling are critical to address the contemporary challenges arising from biomedicine and health sectors, modern industry and the digital economy. The UK Commission for Employment and Skills as well as Tech City UK have identified that a skills shortage in this domain is one of the key challenges facing the UK technology sector: there is a severe lack of trained researchers with the technical skills and, importantly, the ability to translate these skills into effective solutions in collaboration with end-users. Our proposal addresses this need with a cross-disciplinary, cohort-based training programme that will equip the next generation of researchers with cutting-edge methodological toolkits and the experience of external end-user engagement to address a broad variety of real-world problems in fields ranging from mathematical biology to the high-tech sector. Our MSc training (and continued PhD development) will deliver a core of mathematical techniques relevant to all applied modelling, but will also focus on two cross-cutting methodological themes which we consider key to complex multi-scale systems prediction: modelling across spatial and temporal scales; and hybrid modelling integrating complex data and mechanistic models. These themes pervade many areas of active research and will shape mathematical and computational modelling for the coming decades. A core element of the CDT will be productive and impactful engagement with end-users throughout the teaching and research phases. This has been a distinguishing feature of the MathSys CDT and is further expanded in our new proposal. MSc Research Study Groups provide an ideal opportunity for MSc students to experience working in a collaborative environment and for our end-users to become actively involved. All PhD projects are expected to be co-supervised by an external partner, bringing knowledge, data and experience to the modelling of real-world problems; students will normally be expected to spend 2-4 weeks (or longer) with these end-users to better understand the case-specific challenges and motivate their research. The potential renewal of the MathSys CDT has provided us with the opportunity to expand our portfolio of external partners focusing on research challenges in four application areas: Quantitative biomedical research, (A2) Mathematical epidemiology, (A3) Socio-technical systems and (A4) Advanced modelling and optimization of industrial processes. We will retain the one-year MSc followed by three-year PhD format that has been successfully refined through staff experience and student feedback over more than a decade of previous Warwick doctoral training centres. However, both the training and research components of the programme will be thoroughly updated to reflect the evolving technical landscape of applied research and the changing priorities of end-users. At the same time, we have retained the flexibility that allows co-creation of activities with our end-users and allows us to respond to changes in the national and international research environments on an ongoing yearly basis. Students will share a dedicated space, with a lecture theatre and common area based in one of the UK's leading mathematical departments. The space is physically connected to the new Mathematical Sciences building, at the interface of Mathematics, Statistics and Computer Science, and provides a unique location for our interdisciplinary activities.

The increasing demand for low and zero carbon buildings in the UK has provided significant challenges for the energy intensive materials we currently rely on. At present somewhere between 20% and as much as 60% of the carbon footprint of new buildings is attributable to the materials used in construction; this is predicted to rise to over 95% by 2020. If the UK is to meet agreed 80% carbon reduction targets by 2050 it is clear that significant reductions in the embodied carbon of construction materials is required. What also seems clear is that current materials and systems are not capable of delivering these savings. The drive for an 80% reduction in carbon emissions, a decreasing reliance on non-renewal resources and for greater resource efficiency, requires step changes in attitude and approach as well as materials. Improvement in construction systems, capable of providing consistently enhanced levels of performance at a reasonable cost is required. Modern developments in construction materials include: eco-cements and concretes (low carbon binders); various bio-based materials including engineered timber, hemp-lime and insulation products; straw based products; high strength bio-composites; unfired clay products utilising organic stabilisers; environmentally responsive cladding materials; self healing materials; smart materials and proactive monitoring; hygrothermal and phase change materials; coatings for infection control; ultra thin thermally efficient coatings (using nano fillers); ultra high performance concretes; greater use of wastes; and, fibre reinforcement of soils. However, very few of these innovations make the break through to widespread mainstream use and even fewer offer the necessary step change in carbon reductions required A low carbon approach also requires novel solutions to address: whole life costing; end of life (disassembly and reuse); greater use of prefabrication; better life predictions and longer design life; lower waste; improved quality; planned renewal; and greater automation in the construction process. As well as performance, risk from uncertainty and potentially higher costs other important barriers to innovation include: lack of information/demo projects; changing site practices and opposition from commercial competitors offering potentially cheaper solutions.. A recent EPSRC Review has recognised the need for greater innovation in novel materials and novel uses of materials in the built environment. The vision for our network, LIMES.NET, is to create an international multi-disciplinary community of leading researchers, industrialists, policy makers and other stakeholders who share a common vision for the development and adoption of innovative low impact materials and solutions to deliver a more sustainable built environment in the 21st Century. The scope of LIMES.NET will include: adaptive and durable materials and solutions with significantly reduced embodied carbon and energy, based upon sustainable and appropriate use of resources; solutions for retrofitting applications to reduce performance carbon emissions of existing buildings and to minimise waste; climate change resilient and adaptive materials and technologies for retrofitting and new build applications to provide long term sustainable solutions. In recognition of their current adverse impacts and potential for future beneficial impacts, LIMES.NET will focus on bringing together experts to develop pathways to solutions using: renewable (timber and other plant based) construction materials; low-impact geo-based structural materials; cement and concrete based materials; innovative nano-materials and fibre reinforced composites. Through workshops and international visits the network will create a roadmap for multidisciplinary research and development pathways that will lead to high quality large research proposals, and an on-going virtual on-line centre of excellence.

More frequent intense rainfall events, associated with climate change, increase the likelihood of shallow slope failures that lead to costly disruption of road and rail journeys, with risk to life and property. There have been recent slope failures adjacent to transport corridors in the UK, sometimes disrupting important road and rail routes for days. Vegetation has a stabilising effect on slopes: Plant root systems interlock with the soil, increasing its stiffness and strength. Uptake of water by root systems dries the soil profile, again increasing soil stiffness and strength. However, engineers need to be able to predict the combined root reinforcement and soil drying effects on slope stability, so that vegetation management can be used proactively to decrease the probability of slope failure. Vegetation has numerous benefits over conventional hard-engineering solutions, in terms of burying carbon in the soil, enhancing biodiversity, and improving the aesthetic quality of the environment for society. This project will develop and test a quantitative coupled hydro-mechanical model for the in-service and ultimate-failure performance of slopes planted with vegetation. Rooted-soil represents an innovative sustainable construction material, with distinct mechanical and hydrological properties, that can be used in geotechnical systems. The model will be applicable to both slopes covered with natural vegetation and slopes where vegetation and soil have been chosen and managed according to engineering principles. The validated model will provide a clear framework for assessment and remediation of slopes with potential for reducing economic and carbon costs. The model will be developed within a multi-scale continuum modelling framework. It will build on knowledge of the elemental components of the system, working from individual soil-root interaction, to continuum soil-root system, and to complete slope, incorporating spatial variability of materials. Modelling will be informed by X-ray CT imaging of the 3-D deformation of rooted soil undergoing shear, using the micro-VIS facility at the University of Southampton, and by field data from slopes containing established vegetation. Predictions of slope performance will be validated against scaled-slopes in the Dundee geotechnical centrifuge under different rainfall regimes. The geotechnical centrifuge enables the testing and monitoring of small-scale slopes containing roots at realistic stresses, which can be manipulated until the slopes ultimately fail. Template guidelines will be produced for a manual of combinations of plant species, soils and management schemes for optimum performance of designed soil-plant systems suited to emerging climatic conditions. Example data will also be included to allow cost-benefit analyses when designing for slope improvement using vegetation. The potential to translate research findings into related areas will be investigated (e.g. river banks, sand dunes, flood embankments, agricultural and amenity systems). We will engage with an important group of Project Partners, representing key industrial sectors and infrastructure owners, to facilitate the rapid adoption of research findings.