The emerging development of automated driving demands a mutual understanding and a smooth coordination between human driver and vehicle controller, so as to avoid conflict and mismatch in demands, and instead achieve desirable driving performance, smooth and swift transitions which enhance driving safety during complex operating scenarios. However, such driver-vehicle collaboration during automated driving will impact on the driver's attention and cognition and it is important to consider these effects in order to prevent any negative impact on driving. This project aims to achieve a safe engagement and smooth and swift control-authority shift between the driver and the vehicle controller during adaptive automated driving. To this aim, we will first conduct a comprehensive study of driver attention and cognitive control characteristics when interacting with the vehicle controller. An optimal control authority shifting system which considers driver cognition will then be systematically developed and validated. This cross-disciplinary research challenge will be addressed using a unique combination of researchers from engineering, cognitive neuroscience and human factors. The research will not only contribute to the cutting-edge technology innovations in automated driving, but will also result in a major advance in the science of human attention and cognitive control when interacting with automation.

There are two ongoing revolutions in modern automotive industry. The first is the development of autonomous transportation systems which is leading to greatly improved safety, traffic economy, environment and passenger comfort. The second on is the development of advanced propulsion systems, envisaging reduced fuel consumption and exhaust emissions. The intersection of the autonomous transportation systems and advanced propulsion systems is the future trend, which has been revealed in the strategic partnerships between automakers and IT companies. However, there is a big challenge that the environment information collected by autonomous vehicles is poorly used in propulsion systems, especially when vehicles run in fast changing conditions. Only few examples of using environment data to optimise energy efficiency can be found, although the potential benefits in fuel reduction and mission flexibility are great. This project aims to tackle this challenge by developing a cloud-aided learning framework to merge the two themes as integrity. To establish awareness of the environment, the onboard sensors of autonomous vehicles including cameras, light detecting and ranging (LiDAR), ultrasonic, and radar are used to perceive the environment over short distances. The GPS and intelligent transportation systems are used to perceive the environment at a further distance. The combined information enables the autonomous vehicle to establish a comprehensive model of the external environment. Using advanced machine learning algorithms (e.g., dynamic Bayesian network), the environment information can be used to update the propulsion system model in real time. Combining the real-time updated model with dynamic optimisation methods (e.g., adaptive model predictive control), the optimal actions of the propulsion system can be obtained in s systematic way. Employing high performance computing resources on cloud, the computational intensive modelling and optimisation tasks can be cost-effectively addressed. Aligning with the EPSRC Innovation Fellowship priority area of "Robotics and Artificial Intelligence Systems" through its focus on efficient transport, a cloud-in-the-loop testing platform will be built. This framework focuses on sensing, modelling, control, optimisation and computing of energy efficient autonomous vehicles. In a long term vision, the framework can be generalised from a single vehicle to connected and autonomous for further economic benefits.

There are two ongoing revolutions in modern automotive industry. The first is the development of autonomous transportation systems which is leading to greatly improved safety, traffic economy, environment and passenger comfort. The second on is the development of advanced propulsion systems, envisaging reduced fuel consumption and exhaust emissions. The intersection of the autonomous transportation systems and advanced propulsion systems is the future trend, which has been revealed in the strategic partnerships between automakers and IT companies. However, there is a big challenge that the environment information collected by autonomous vehicles is poorly used in propulsion systems, especially when vehicles run in fast changing conditions. Only few examples of using environment data to optimise energy efficiency can be found, although the potential benefits in fuel reduction and mission flexibility are great. This project aims to tackle this challenge by developing a cloud-aided learning framework to merge the two themes as integrity. To establish awareness of the environment, the onboard sensors of autonomous vehicles including cameras, light detecting and ranging (LiDAR), ultrasonic, and radar are used to perceive the environment over short distances. The GPS and intelligent transportation systems are used to perceive the environment at a further distance. The combined information enables the autonomous vehicle to establish a comprehensive model of the external environment. Using advanced machine learning algorithms (e.g., dynamic Bayesian network), the environment information can be used to update the propulsion system model in real time. Combining the real-time updated model with dynamic optimisation methods (e.g., adaptive model predictive control), the optimal actions of the propulsion system can be obtained in s systematic way. Employing high performance computing resources on cloud, the computational intensive modelling and optimisation tasks can be cost-effectively addressed. Aligning with the EPSRC Innovation Fellowship priority area of "Robotics and Artificial Intelligence Systems" through its focus on efficient transport, a cloud-in-the-loop testing platform will be built. This framework focuses on sensing, modelling, control, optimisation and computing of energy efficient autonomous vehicles. In a long term vision, the framework can be generalised from a single vehicle to connected and autonomous for further economic benefits.

Control systems play a central role in automation and modern industry. By using feedback, control systems are designed to regulate system outputs around a reference setpoint or track a reference trajectory, despite disturbances and variations due to its operational environment; for example, an aircraft follows a specified speed and altitude despite gust, wind and changes typical of an aircraft (e.g. number of passengers). Performance specifications for control systems are typically defined to quantify their behaviours in terms of the defined reference. The analysis and design tools/methods across the entire area of control engineering are mainly developed and built upon these specifications. With the demand for ever increasing levels of automation, we are moving towards goal-oriented operation, where what a system needs to achieve is specified at a high level (e.g. in terms of economic or mission requirements), rather than how it is to be achieved (e.g. through defining a setpoint or trajectory). The goal-oriented operation improves the operation process, offers new opportunities for technological advances and reduces operational costs. A Goal-Oriented Control System (GOCS) is essential to enable this type of operation. Another significant difference in GOCS is the influence of disturbance/uncertainties on design specifications. Traditionally a central role of a control system is to attenuate the influence of external disturbance/uncertainties because they always divert the system away from a reference. However, certain disturbance may be good for the system operation in terms of a high-level goal. For example, changes in raw material or environment may make a chemical process more profitable so they should not be rejected; a favourable change of wind condition should be exploited, rather than rejected in an emergency landing as it helps the aircraft to glide longer, providing a larger safe margin in performing forced landing. In current control system design, constraints (e.g. physical, operational, legal) are often considered explicitly or implicitly through generating appropriate references. However, as the system specifications will be defined at a high level in goal-oriented operation and the control system must work out how to achieve the specifications/goals, constraints must be specified very carefully in order to meet safety and other requirements. Some constraints are difficult to represent in the specifications or to take into account constraints within current control design. This Fellowship will develop analysis and design tools for goal-oriented control systems, build up a unified framework for the next generation of control systems in performance specifications, constraint representation and problem formulating, investigate its performance and other properties in the presence of disturbance, uncertainties and constraints, and benchmark their applications. Temporal logic will be used to represent high level specifications and a wide range of complex constraints. By combining temporal logic with rich representation of dynamic systems in control engineering, it aims to develop a complete new analysis and design framework for GOCS that involves both rich dynamics and complex specifications in performance and constraints, and to provide analysis tools for the propagation of information errors, environmental disturbances and dynamic uncertainties into the high level performance specifications and fulfilment of constraints, and the interplay of these terms with a feedback strategy and controlled dynamics. The Fellowship programme is developed and built upon the Fellow's internationally leading work in disturbance observer-based control, model predictive control and autonomous vehicles in the last 25 years. The successful completion of the Fellowship will open a new field of goal-oriented control systems, transform control engineering and unlock the potential of moving from low levels to high levels of automation

In recent years, a lot of research has been carried out in the field of energy management for full HEVs and EVs. Strategies that are based on heuristics can be easily implemented in a real vehicle by using a rule-based strategy or by using fuzzy logic. To find the global optimal solution, control techniques such as linear programming, quadratic programming, optimal control, especially dynamic programming have been studied. A different approach has been proposed in some recent work. In this approach instead of considering one particular driving cycle for calculating an optimal control law, a set of driving cycles is considered, resulting in a stochastic optimization approach. After considering the optimisation based methods described above, the following observation is made. The principal common drawback of all the aforementioned strategies is consideration of drivability as an afterthought. The drivability is considered in an ad hoc fashion as these approaches are not dynamic model based. At best, techniques such as game-theoretic optimisation utilises quasi-static models which are not sufficient to address drivability requirements. Another important drawback of these strategies is robustness. Theses strategies do not include a feedback control block, so the robust performance of the strategy is not guaranteed as the vehicle parameters deviated from their nominal conditions. Ad hoc adaptation of these strategies for drivability does impact their optimality and therefore a negative impact on the emissions and fuel consumption. The robust multivariable control has extensively been used in the aerospace industry, process control and chemical and petrochemical plants. However, the application of the multivariable control in the automotive industry is scarce and when it comes to the energy management development, there is no application. There exists multivariable based method such as Model Predictive Controls, however, these techniques are mainly time domain based, which has its drawbacks and they have the limitations such as on-line implementation is not possible. The multivariable control design has been used by this author for integration of active chassis systems (Active Roll Control and Active Limited Slip Differential). There is a tremendous opportunity for reducing fuel consumption, emissions, and calibration effort to production utilising this methodology. The user fuel economy data obtained from Toyota Prius has shown that there is approximately 10-15% difference between the average real world and driving cycle fuel economy. This difference is mainly due to the drivability impact. Our conservative projection of fuel savings and CO2 reduction of this proposal is 5-7%. An approximate annual savings of 400 Million, due to reduction in calibration effort, based on a volume projection of 50,000 hybrid vehicles after the end of this project. Furthermore, additional possible benefits and savings are foreseeable due to transferability between platform variants. The aim of this work is to design and develop a multivariable feedback controller to replace the ad hoc design currently used to address robustness and vehicle drivability issues as stated above. The robust feedback multivariable is derived based on the dynamic models of the plant so the drivability requirements are addressed as a part of the design and development of the controller and it is not an afterthought. Furthermore, the lack of robustness is not an issue with the Multivariable controls as they are inherently derived from feedback policies.