| |
| Sensible Solutions - evolving sensor technology in ADAS. |
18/10/2004 |
| What chance have the Swedes - or anyone else - of achieving a zero tolerance of fatal accidents? Leaving aside those reckless souls for whom no amount of education or technological assistance is likely to make a scrap of difference, by how much can the various ADAS (Advanced Driver Assist Systems) projects improve safety on the roads? And at what price? At what stage does technology risk crossing into the tangled web of driver responsibility, of running foul of an increasingly litigious community? Is there, in other words, a fail-safe means for technology to assist in the process of saving lives?
Governments around the world have spent countless millions on advertising programmes and enforcement initiatives designed to change driver behaviour for the better. And there is no doubt that such initiatives have had their effect. Yet equally, there is evidence to suggest that at least some of the improvements in the KSI (killed or seriously injured) rate has had as much to do with advances in medical science as any alteration in the long-term behaviour of drivers. It has been this recognition, perhaps more than any other factor, that has forced the pace of research into road and vehicle safety systems as a means of achieving the ambitious casualty reduction targets set by the European Union, the United States and others.
One of the more spectacular advances has been in the field of sensors which, from their original appearance in adaptive cruise control systems, now form a part of most research and development work, including a number of high-profile programmes such as the IVI (Intelligent Vehicle Initiative) in the US, the AHS (Advanced Highway System) and the ASV (Advanced Safety Vehicle) in Japan and a range of R & D projects in Europe, including the 6th Framework Programme's E-SAFETY initiative. All these make extensive use of computer, wireless and camera technologies to improve the range and quality of sensor applications.
"This technology has a lot of future in safety-related research," said Olivier Aycard, Associate Professor at the Joseph Fourier University (UJF) in France and project manager of the ProFusion Project looking into sensor fusion. "Most up-coming systems are using sensors in one form or another."
Moving rapidly beyond the confines of ACC (Adaptive Cruise Control), the sensors now use a range of communications protocols such as microwave infrared, ladar and radar to detect and, in the future, classify up-coming objects in the carriageway. Applications currently being worked on vary from collision avoidance, through lane keeping/lane support, to pre-crash preparation, obstacle classification, rear detection, driver drowsiness and much more.
Amongst the work being done is research from a consortium led by e2v Technologies in England and including BAE Systems, LEW Techniques of Taunton, Somerset and the University of Birmingham, which is attempting to combine both long range and short range applications within a single sensor, despite the enormous difficulties inherent in such a project.
In Germany, Siemens are developing a device that makes use of light to accurately measure the distance between a vehicle and another object using a CMOS (Complementary Metal-Oxide Semiconductor) image sensor chip manufactured at the Fraunhofer Institute IMS. By utilising the ultra-fast, millionth of a second aperture (shutter) speeds available through CMOS, measurement is made of the time taken for a laser pulse to reach and return from an object.
"Using a CMOS image processor allows for an extremely accurate measurement and means that we can obtain a 3D image of an object," said Dr Peter Mengel, a Principle Research Scientist with Siemens and Project Manager for the UserCam Project. "We are testing it at distances of between 20 and 30 metres, the distance at which optics gives a better resolution than radar, and intend it for use in urban driving conditions."
Yet as Dr Mengel is the first to admit, the use of light has its limitations particularly as regards the amount of processing power required to classify the images but also in dealing with conditions of adverse weather and lighting. The use of optics in urban driving conditions has, according to Dr Eddie Moxey, head of the Mechatronic Systems and Robotics Research Group at the University of Surrey, England, some way to go.
"… processing image data," said Moxey, "even from low resolution can be time consuming especially when tracking shapes."
But even if the use of light were to be rejected in favour of radio signals, a number of further problems arise. Radio signals are the medium of choice for ACC applications but there is no currently available sensor capable of dealing with both long and short range applications. And the deployment of two sensors in place of one is not, according to Nigel Priestly, Chief Engineer at the Chelmsford, England, headquartered e2v Technologies, sustainable.
"Vehicle manufacturers are running out of space from which to hang bits of equipment," said Priestly. "If we can develop a single sensor that does both jobs, it would make sense to do so."
The incentive to do so is not hard to find. By 2012, regulations within the EU will require the migration of the short-range, wide angle sensors operating at the 24 GHz wave-band to 72-77 GHz, the same frequency occupied by the narrow angle sensors used for long-range work. The decision about the nature of forward facing sensors has, therefore, already been made and it is for the ITS industry to try and resolve matters.
"We do not want to be in the position of being pushed by the regulations," said Priestly, "so we have begun looking for a solution that will see a single sensor using radio-waves to deal with both long and short-range functions. What it amounts to is that all research work on sensors at 24 GHz will be rendered useless by the new regulations."
In Europe, the timetable for the development of safety-related applications for sensors is being driven by the political objective of a 50% reduction in the KSI rate by 2010. For this reason, the horizons of the EU co-funded projects like SafeLane, UserCam, Profusion, Safespeed, Compose and others is limited to the expansion of existing technologies rather than the development of new systems.
"Our first target is to produce the technology for use in a test vehicle in two to three years," said Dr Mengel of Siemens. "Active intercession is not part of our remit although OEM partners may want to consider this in the future. We are simply upgrading technology first developed under MissI, a 5th Framework Programme project."
Echoing this view, Dr. Angelos Amditis, a Research Associate Professor at the ICCS- NTUA Microwaves and Optics Lab in Athens said, "Developing new technology takes a long time so we have to make more effective, cleverer use of existing sensors. Fusion is important in this regard as is the use of more powerful processors, better lens and the development of an integrated platform that is capable of providing an image and information of what is happening all around the vehicle. But don't expect to see such a fully commercial new platform by 2010."
Nevertheless progress is being made. Lateral Safe, a project being jointly managed by Dr Amditis and Dr Kaiser Diechoff of Bosch, as a sub-set of the PReVENT IP project (6th FW, IST), is looking at the extent to which signal processing, especially at the area of lateral monitoring and warning can be improved. Amditis sees accelerating computer technology and signal processing techniques as the key to the complex issue of identifying objects with the degree of certainty necessary for safe driving.
"Sensor fusion," said Amditis, "helps to overcome the identification disadvantages of stand-alone sensors. It is the El Dorado of the new generation of ADAS."
Overlying these developments is, for vehicle manufacturers, the ever present threat of litigation arising out of the provision of active ADAS systems and it would be idle to pretend that such concerns play no part in the decision making process. The impending migration of the radio frequences required by the EU post 2012 is likely to render obsolete all the research into 24 GHz sensors and put back the search for a viable short range solution. Using CMOS technology presents its own set of difficulties including the time taken to process the images and make sense of them. In general terms there are serious doubts over the ability of such technology to deal with multiple targets at a fast enough rate to work, even at the slow speeds of normal urban traffic.
"There simply isn't enough time to perform complex processing, especially in an urban environment where both the number of potential targets and the level of background noise and clutter is very high," said Dr Moxey. "The new 3D images created from new CMOS and Lidar technology produce fast images but I am not convinced that we will be able to use them to track multiple targets in real-time. It is a hugely difficult area."
Should, then, reliance be placed on sensors and other ITS developments to deliver the reductions in road casualties sought by politicians? One of the strongest messages to come out of the research for this article was that development in this area is, by its nature, a slow process. The evolving 2nd generation sensors upon which so much hope is vested, have a considerable way to go before they can be deployed in production line vehicles. When the concerns over litigation are added, there seems little prospect of seeing the wholesale introduction of safety-related equipment, particularly those active devices which have the most to offer but which are, at the same time, the legal equivalent of a quadmire. The optimistic view sees the introduction of such systems in around five years. Certainly the ability of the technology to deliver a fail-safe means of saving lives will be that much more advanced by 2010 but it is not technology which is currently holding up the pace, it is the concerns of the vehicle manufacturers. And who can blame them for that?
|
For further information, contact :
1. Dr Peter Mengel Siemens AG, Corporate Technology CT PS 9Otto-Hahn-Ring 6, D-
81739 Munichon, Tel: #49 89 636 48756, E-mail: peter.mengel@siemens.com
2. Olivier Aycard GRAVIR-IMAG & INRIA Rhône-Alpeson 655 avenue de l'Europe -
Montbonnot,38334 Saint Ismier, Cedex. Tel: #33 4.76.61.55.14, E-mail: olivier.aycard@imaq.fr
3. Nigel Priestley of e2v Technologies, Cchelmsford, England. Tel: #44 1522 585 458, e-mail:
nigel.priestley@e2v.com
4. Dr Angelos Amditis, ICCS- NTUA Microwaves and Optics Lab, 9, Iroon Politechniou str., Polytechnic Campus - DECE - office 2131 15773 Zografou, Athens, Greece, Tel: #30 210 7722398, e-mail: aamditis@mail.ntua.gr
5. Dr Eddie Moxey of the University of Surrey, England. Tel: #44 1483 300 800, e-mail: e.moxey@surrey.ac.uk
|
|
