Keep a steady speed. Anticipate what’s ahead. Where possible, coast to slow down. When accelerating, do it gently. Driving tips like these don’t just reduce your fuel usage and save you money – they also cut carbon emissions and help tackle climate change. It’s win-win.

But is boosting fuel efficiency on our roads simply a question of us becoming more ‘savvy’ when we’re behind the wheel, or do cars need to get smarter too?

In fact, both have a big role to play. That’s the starting point for G-Active, a collaboration between SES members the University of Southampton and Imperial College London. The aim? To develop an on-board system that predicts traffic conditions and the driver’s own behaviour, harnessing this information to optimise the car’s energy use and provide the driver with real-time fuel efficiency advice. The ultimate goal? Significant inroads into the transport sector’s carbon footprint.    

Drivers of Change

“Improving fuel efficiency on our roads isn’t just about coming up with a new car engine or a revolutionary vehicle design,” says Roberto Lot, Professor of Automotive Engineering at Southampton. “G-Active is focusing on the here and now. Our system aims to make existing vehicles more intelligent, and to work with drivers to modify the way they currently drive.”

“Can an engineering system really modify driver behaviour and deliver better fuel efficiency on our roads?”

– Dr Simos Evangelou, Imperial College London

The stakes are high. Transport produces 23% of UK greenhouse gas emissions, with cars and taxis accounting for over half of this percentage.[1] For G-Active (or ‘Green Adaptive Control for Future Interconnected Vehicles’, to give its full title) the target is to enable cars and light-duty vehicles to achieve a cut of over 5% in fuel use and emissions. As part of this 3-year EPSRC-funded project, Southampton and Imperial will test and validate their ideas in conjunction with a leading UK car-maker, providing a route to real-world application.

Fundamentally, driving is all about the interaction between humans and machines. G-Active is recalibrating that relationship by exploiting the Data Revolution.

“We aim to make existing vehicles more intelligent and work with drivers to modify the way they drive”.

– Professor Roberto Lot, University of Southampton

“Today, how fast and smoothly we drive is determined by the way we perceive and react to what’s around us, and by how we feel,” Professor Lot explains. “That’s not a very efficient way of doing it. It would be far better if the vehicle itself could accurately predict how much power will be required for the journey ahead, and then produces that power in as smooth a way as possible. To do that, it needs detailed data – such as how fast the car is going, where exactly it is, what’s ahead, how it’s being driven and what the weather’s like. Even half a minute of future estimation will help to reduce fuel consumption.”

Dr Simos Evangelou of Imperial is a key member of the G-Active team: “In fuel efficiency terms, it’s hard for a human driver to pick the optimum moment to start braking or accelerating”, he says. “Technology can do it much better. If we’re to achieve success, though, we need to bring many perspectives to this challenge. Electrical, mechanical and control engineering and a variety of high-level computer programming skills have a part to play. But so does expertise in human factors and the design of human-machine interfaces. Pooling the skills of our two institutions will enable us to meet these complex needs.”

Fuelled by Data

Designed for use on conventional, electric or hybrid cars, the G-Active energy management system will incorporate three elements:

  • Perception Layer’ – gathering data and processing it using a bespoke computer model to produce short-term predictions of driver behaviour and power needs.
  • ‘Decision Layer’ – using bespoke software to analyse these predictions and select the most appropriate speed for the vehicle.
  • ‘Action Layer’ – putting decisions into effect by sending instructions to the vehicle’s drive train.

Data underpins everything,” Roberto Lot says. “Our approach is to harvest what’s readily available. Speed and acceleration data can be collected from the in-wheel sensors already fitted to cars; we can piggy-back GPS navigation systems to pinpoint the car’s location; we can use SatNavs to obtain route information; and dashboard cameras or electromagnetic reflectors can be used to calculate the distance to the next vehicle.”

Future possibilities include sharing data between vehicles, harvesting data from internet-connected road infrastructure (e.g. indicating whether there’s congestion at traffic lights ahead) and tapping into online traffic-prediction tools.

Overall, the G-Active system will gather substantial volumes of data for processing in real time. Most will soon be deleted as it becomes irrelevant to the vehicle’s needs, but some aggregated data will be stored as a source of information on the driver’s behaviour and preferences, which can then be factored into the system’s decision-making.

The outcome will be a system that thinks ahead on an ‘ever-receding horizon’ basis and offers three modes of use:

  • Standard’ – the driver’s intentions are predicted and, within these constraints, the system subtly optimises energy management.
  • ‘Coaching’ – the driver receives recommendations on fuel-efficient driving behaviours that suit current traffic conditions.
  • ‘Autonomous’ – energy management and vehicle speed are optimised without any reference to the driver.

After concepts have been tested on the Southampton University Driving Simulator, a prototype will be trialled by G-Active’s industrial partner. “We’re keen to bring more partners on board as the project progresses,” Dr Evangelou adds.

The Human Touch

So much for the technology. But what about the driver? Will they resent a ‘back-seat driver’ that actually affects the way their vehicle performs?

We need to ensure that drivers want to use the system and don’t find it distracting or annoying,” says Roberto Lot. “So it’ll be designed to be able to take their individual preferences and driving style into account. Suggestions to the way they drive will be made discreetly and unobtrusively.”

The team are currently developing dashboards displaying a range of recommended speeds that vary with traffic conditions. Easy to understand and similar to dashboards already familiar to drivers, this is the sort of modest change that could maximise the system’s acceptability.

“The big question is, can an engineering system really modify driver behaviour and deliver better fuel efficiency on our roads?” Simos Evangelou concludes. “We’re confident that G-Active can achieve this. It’s a case of smarter car, smarter driver.”

Visit the G-Active Website

Visit the website

Project Contacts

Prof. Roberto Lot

Prof. Roberto Lot

Faculty of Engineering and the Environment

University of Southampton

Dr. Simos Evangelou

Dr. Simos Evangelou

Department of Electrical and Electronic Engineering

Imperial College London

Further Information

Footnotes

[1] Source: Department for Transport, 2016.

Image Captions

Header Image:  Head model reconstructed from full-head, high-resolution MRI images (Human Connectome Project, Study ID: MR20170307185242 Subject ID: F3T_2015_16_056).

Image 2: VPH-DARE@IT is delivering new insights vital to accelerating dementia diagnosis.

Publications

Please visit the G-Active website

Leave a Reply