In 2013, more than four people died on the UK’s roads every single day – a total of 1,713 fatalities. The official statistics for that year also showed that more than 20,000 people suffered serious injuries in road accidents. These grim facts are the biggest motivation for the ongoing development of intelligent systems to make cars safer. All major automotive manufacturers and their suppliers, in addition to Google, hi-tech firms and telecoms companies, are making strides towards the vision of autonomous – or self-driving – cars. Significant amounts of funding have been invested by national governments, the European Union and the defence sector in the past two decades, with no sign of their commitment waning.
The quest towards the Holy Grail of a mass-produced, fully autonomous car is the fourth in a series of revolutions in automotive safety. The first, spanning the 1950s to the 1970s, saw the introduction of the modern seat belt and airbags, with the aim of reducing the impact of accidents on passengers. The second revolution, up to the 1990s, involved developments in active safety systems, such as anti-lock brakes, that tried to avoid, or at least minimise, the severity of accidents for all road users. The third revolution occurred in the late 1990s and saw active safety systems intervene in the driving process, expanding the sensing range beyond the car’s bumpers.
By the start of the current decade, the option in ordinary family cars for radar sensors to control the brakes and accelerator as part of adaptive cruise control systems had become commonplace. We are now entering the fourth revolution, where advanced sensors, communications and computational intelligence are all combining to reduce the societal cost of road transport by anticipating and avoiding accidents, minimising the environmental footprint of road transport and further reducing road casualties.
What drivers want, and what drivers need, will shape the future of automotive technology. Recent surveys indicate that most drivers would like to retain some form of control over their car and would accept fully autonomous driving only under very specific conditions, such as long motorway journeys. On the other hand, drivers would welcome technologies that reduce their workload and intervene proactively when fatigue or lapses of concentration happen.
The eyes, ears and ‘feel’ of any such future systems would come from its sensors. These are both inward-looking (at the driver and the vehicle itself) and outward-looking (at the environment). In terms of development, the tougher sensors to invent are the outward-looking ones: they have to be able to tell if you are driving close to a pedestrian, a cyclist or another car, as well as whether you are driving on-road or off-road. They even possess situational awareness: the car needs to figure out whether swerving to avoid an unexpected obstacle will lead to an uncontrolled skid, or whether it can avoid the accident altogether. The cameras involved are high-resolution and are able to provide 3D images, but sadly they do not work when it is foggy or in dense spray. They may also get blinded by the sun or even headlamps. Acoustic sensors are great for parking and for other very short-range applications, but once again they cannot be trusted in noisy environments, at high speed, or operating over distances above 100 metres.
Research at the University of Birmingham – funded by the automotive industry and government – is making significant strides towards creating sensors that work in all weather conditions and do not have these drawbacks. Radar sensors suffer from none of these limitations, and their use in an automotive context was pioneered by researchers in Birmingham’s School of Electronic, Electrical and Systems Engineering as far back as the 1980s. Our current radar research aims to produce images with optical quality and resolution at high vehicle speeds and in all environments and weather conditions. Operating frequencies in thus far underexploited parts of the electromagnetic spectrum, such as the bands used in airport scanners, are being explored, in addition to exploiting all properties of the radar signals with advanced signal processing techniques. This research is helping us move closer towards the vision of autonomous cars.
Dr Peter Gardner
Head of School of Electronic, Electrical and Systems Engineering and Reader in Microwave Engineering, University of Birmingham
Dr Edward Hoare
Senior Research Fellow in Automotive Sensors, University of Birmingham