Green adaptive control for future interconnected vehicles

Lead Research Organisation: University of Southampton
Department Name: Faculty of Engineering & the Environment

Abstract

Vehicle energy management (EM) systems currently concentrate on controlling the drivetrain to deliver the requested power to the wheels optimally from one or more energy sources, depending on the level of hybridisation of the drivetrain. Despite the existence of a vast range of such systems, encompassing rule-based to optimisation-based schemes, a number of challenges remain and opportunities exist to realise the next generation of more efficient EM control. The Green Adaptive Control for Future Interconnected Vehicles project aims to directly address these challenges by developing, implementing and testing EM systems that will now be global (simultaneous optimisation of the drivetrain energy, auxiliary systems energy and driving speed rather than only of the drivetrain energy), predictive (optimisation over a 'look ahead' horizon rather than just based on the instantaneous power demand), and newly adaptive (taking into account driver's preferences, traffic and other environmental conditions). The ultimate goal is to reduce by more than 3-5% the fuel consumption of the future fleet of passengers and light duty vehicles for a range of drivetrain architectures (conventional, electric and hybrid electric) and auxiliary systems (cooling systems, and other). To reach this objective this project will design, implement and demonstrate a new generation of EM together with an Adaptive Cruise Control system, which will automatically drive the vehicle at the most appropriate speed. For this to be effective, we also need to make the drivers aware of the benefits and to make small changes in their driving behaviour. Indeed, substantial reductions in energy consumption can be achieved by making small changes to the behaviour of a large number of drivers. Human factors methods will be used in this research to optimise the design of such new EM control systems.

The proposed EM systems will have three operating modes: Autonomous, Coaching and Manual, which are all based on the same three layers structure. The first one is the Perception layer, which has the purpose of gathering navigation (e.g. route) information, driving information (e.g. the vehicle position, speed and acceleration), information related to the surrounding vehicles, and finally infrastructure conditions (e.g. the state of the next traffic lights series). We will use this information to feed the Decision layer, which is where the intelligence of the system will lay, and which will also be the core of our project. In the Autonomous mode, the system will manage the car in a much smarter way than a human driver by selecting, case by case, the most appropriate vehicle speed and acceleration taking into account all environmental constraints such as road characteristics, desired time to destination and traffic conditions. Once the EM and speed will be optimised, the Action layer will safely drive the vehicle at the most appropriate speed thanks to the Adaptive Cruise Control system. Even if drivers are not always keen to accept such autonomous systems and want to drive according to their personal style, significant fuel reduction may be achieved by using predictive optimisation, in which the system tries to anticipate the future power demand, which is predicted by the system itself according to the information available. Indeed, by selecting the Manual operating mode, the driver behaviour will be predicted by using a mathematical model that will be appositely developed in this project and eventually we will use such prediction to optimise the EM and reduce fuel consumption. Finally, while using the Coaching operating mode, the most appropriate speed will be calculated by the system and then recommended to the driver by using an appropriate haptic (and possibly visual and acoustic) Human Machine Interface, but the driver will maintain the freedom and the responsibility of keeping the preferred speed.

Planned Impact

It is well expected that road transport greenhouse gas emissions can be reduced by improving further the efficiency of conventional vehicles, increasing the penetration of electric and hybrid electric vehicles, and changing driver behaviour. These changes will be enabled by technological improvements and cost reductions, in which a key role will be played by innovation.

The proposed research will develop the next generation of energy management (EM) control and intelligent cruise control technologies to reduce the fuel consumption of the fleet of future passenger and light duty vehicles by more than 3-5%. This will be achieved by improving the efficiency of conventional, electric and hybrid vehicles, and by providing systems that coach drivers to adopt eco-friendly driving behaviours. These advances will result in a reduction of emissions as well as a reduction of fuel and running costs of automotive transport (3-5% fuel savings expected for traditional drivetrains and 8-10% for hybrid electric drivetrains, corresponding to financial savings of £425/vehicle thus savings up to £100M at national level for every 5% reduction in hybrid drivetrains). The new EM design tools and algorithms will benefit automotive manufacturers (primarily the industrial partner of this project, Jaguar Land Rover) by enriching their existing toolset and procedures for EM control design. Potentially, these innovations will also be usable by the automotive sector of other drivertrains not covered in the present research, such as fuel cell (hydrogen) vehicles, in which similar control and optimisation issues exist. The overall automotive industry in the UK will take advantage of the produced fuel reduction enabling technologies, which will help to increase its technological leadership position at the world stage. Furthermore, the reduction of carbon emissions will enable the Government and industry to move closer to the targets set for the 2020 horizon and beyond, towards 2050. At the same me, the general public will benefit from the availability of more fuel-efficient vehicles, immediately in fuel cost savings, and in terms of public health in the long term. Finally, the research will help to increase the safety of driving, hence accidents will be reduced, and therefore road users and pedestrians will benefit.

The most effective route to reach all the users that may benefit from this research, hence to achieve the maximum impact, is: first, to inform and familiarise the automotive industry about the research progress; second, to encourage and support the implementation of the proposed innovative technologies to actual vehicles; third, to validate and demonstrate the advantages of these technologies; and fourth, the automotive industry to make the proposed innovations available to their customers and the general public. To traverse this route, the project is built on a strong collaboration with one of the key UK automotive manufacturers, Jaguar Land Rover (JLR), who will take part in all the stages of the project. Thus, there will be a privileged closed-loop channel that will allow conveying the research outcomes to their natural industrial application and also it will be communicating to the research team the point of view of the industry. Significantly, some of the versions of the proposed adaptive green control algorithm will be implemented and tested on JLR test vehicles, which is a significant step towards future industrial application of the research. More specifically, the overall support JLR will offer to the project includes the provision of engineering resource, test vehicles and facilities, and sharing of engineering knowledge and expertise. The full support and substantial commitment of JLR signifies the expected benefits and impact that will be achieved by the project.

Publications

10 25 50
 
Title Eco-Driving Habits Questionnaire 
Description We have developed an extensive eco-driving questionnaire, examining respondents current driving habits and motivations to drive in a more fuel efficient way. The survey consists of 7 main sections, focusing on participants 1) Demographics, 2) Current Vehicle Details, 3) Driving Habits, 4) Eco-Driving Motivations and Strategies 5) Eco-Driving Behaviors 6) Future Vehicle Technologies, & 7) Environmental Concern. The Questionnaire, lasts approximately 30 minutes and provides a clear picture of drivers current actions, and motivations touching in the future. 
Type Of Material Improvements to research infrastructure 
Year Produced 2017 
Provided To Others? No  
Impact Currently we have had over 250 participants partially complete the survey. Preliminary results of the questionnaire has been shared at an International conference, and work is on-going to develop this into a full journal article and share the questionnaire with the wider academic community 
 
Title Automotive Data Acquisition Module (ADAM) 
Description ADAM is a low-cost device for unobtrusively collecting naturalistic driving data, suited for small-scale studies designed to analyse certain aspects of driver behaviour. It combines stereo video cameras, a GPS receiver, accelerometers and gyroscopic sensors and can collect several hours of driving data. It is based on two Raspberry Pi single-board computers, and attaches to the windscreen of a vehicle using a suction cup. Software has been developed in Python for data collection. Importantly, ADAM can be given to a study participant to install in their own vehicle, so that it may collect data while they carry out their normal day-to-day driving. The collection of stereo video data allows for many useful quantities to be calculated, such as the spacing between vehicles. 
Type Of Technology Detection Devices 
Year Produced 2017 
Impact ADAM has been used internally within the research project to collect naturalistic driving data, primarily to evaluate natural car-following and cornering behaviour and develop models of driver behaviour. ADAM was presented at the 15th European Automotive Congress in Madrid, 2017, where it gathered some interest, and technical details of the device were given in the associated conference paper. We have encouraged the academic community to implement similar devices, enabling other targeted, small-scale naturalistic studies. 
 
Title Naturalistic Driving Data Processing Software 
Description Software to process naturalistic driving data, such as that collected by ADAM. This has been developed in MATLAB and includes: - Processing of GPS data to provide smoothed speed, heading and path curvature estimates. - Filtering/smoothing of accelerometer and gyroscope time-series data. - Processing of stereo video data to provide robust estimates of range to other vehicles. 
Type Of Technology Webtool/Application 
Year Produced 2017 
Impact Within the project, the software has enabled us to process naturalistic driving data from ADAM and develop models of driver behaviour. Details of the software were presented at the 15th European Automotive Congress in Madrid, 2017, and we have encouraged the academic community to implement similar systems to enable other targeted naturalistic driving studies. 
 
Description workshop on Eco-Driving and Energy Management Optimisation 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Study participants or study members
Results and Impact 20 people attended the workshop, including researcher from other UK universities and members of the automotive industry (e.g. Ford, Komatsu). The workshop included a series of presentations on eco-driving and energy management optimisation and has been followed by a 45 mins discussion, which highlighted different points of view on the subject and sparked new ideas too. The event was streamed on the web too.
Year(s) Of Engagement Activity 2018