Car makers launch new generations of night vision systems

Citation

(2007), "Car makers launch new generations of night vision systems", Sensor Review, Vol. 27 No. 4. https://doi.org/10.1108/sr.2007.08727daf.004

Car makers launch new generations of night vision systems

Car makers launch new generations of night vision systems

An application that appeared poised to boost dramatically the market for thermal imaging was vehicle night vision systems. When first introduced by Cadillac on its Deville model in 2000 it was heralded as a major breakthrough in road safety, as according to a US Department of Transportation study, more than 20 per cent of fatal accidents occurred between midnight and 6.00 a.m., a period which accounts for only 2.4 per cent of traffic volume.

Initial customer reaction was encouraging with close to 20 per cent of buyers of the 2000 model-year Deville taking up the $2,250 option but this figure subsequently fell to only around 2 per cent.

The Lexus system, launched in 2002, was less costly at around $1,750 but saw a similar decline in sales following an initial flurry of interest. In addition to cost, this reflected limitations in the technology as it then stood, such as problems with false alarms, but now two major European car manufacturers, BMW and Mercedes-Benz, have launched improved systems.

These are based on two differing technologies: far-infrared (FIR) and near-infrared (NIR). The FIR system is passive, detecting the thermal radiation emitted by warm objects in the wavelength range 8-12mm whereas NIR systems are active and use an IR source to transmit light with a wavelength of around 800nm at the object and then detect the reflected signal.

The main advantage of NIR systems is that costs are lower as the sensors used at these wavelengths (CCDs) are already well developed for other imaging applications such as video cameras. The down-side is that they have a shorter operating range.

BMW introduced the FIR-based “Night Vision” system, designed and made by component manufacturer Autoliv, as an optional feature, priced at around $2,000, in September 2005. An image of the road ahead is presented to the driver by an internal display integrated into the dashboard and warm objects such as people and animals show up as bright white features. The system has a range of 300m (984ft), a 368 field of view in the horizontal plane and a refresh rate of 30Hz. The heart of the system is a custom-designed microbolometer – a thermal imaging sensor featuring a 320 x 240 element array that can detect very small temperature differences, typically less than 0.1°C. The sensor is mounted on the car's front bumper just below and to one side of the number plate (Figures 5 and 6). Aside from the microbolometer, the optical elements represent the main cost of the technology.

Figure 5 The infrared detector module used by BMW

Figure 6 The IR detector is the small pale circle located below and to one side of the numberplate

When designing the system, Autoliv investigated both germanium and the chalcogenide glass “GASIR” a compound of germanium, arsenic and selenium (Ge₂₂As₂₀Se₅₈) as potential lens materials. Germanium is a high performance material for long wavelength IR lenses with high mechanical resistance and a refractive index of 4.0. It is, however, expensive and requires a costly process, single- point diamond turning, to machine it into an aspherical surface.

Conversely, GASIR is a new type of IR-transmitting glass with a refractive index of 2.5 and is an attractive candidate for high volume IR lenses as it can be moulded into a finished lens. In addition, the highly variable germanium market price affects GASIR to a lesser extent because it contains only about one quarter germanium. Further, GASIR, unlike germanium, has a refractive index that does not vary greatly with temperature, allowing a simple opto-mechanical design that eliminates the need for active thermal compensation.

Developed jointly with Bosch, the Mercedes NIR “Night View Assist” system is offered as an extra-cost option on the S-Class saloon range. It has two IR sources located within the headlamps which illuminate the road ahead. A CMOS-based IR sensor behind the windscreen detects the reflected light and a grey-scale image is produced on a dashboard-mounted screen (Figure 7). The system has a range of 150m (492ft) which is shorter than BMW's but has better resolution and the images produced are crisper and clearer. Whilst it cannot specifically highlight heat- generating living objects such as animals, cyclists or pedestrians, it responds well to cold objects or those at the same temperature as their surroundings which cannot be picked up by an FIR system, such as obstacles in the road or dead animals. A further potential benefit is that NIR hardware can, in principle, be combined with other safety functions such as lane departure warning systems, a development that is anticipated in the future.

Figure 7 The dashboard-mounted display in the Mercedes NIR system

The latest FIR systems are able to detect pedestrians from longer distances than their NIR counterparts but they are more expensive to manufacture; NIR detectors offer superior resolution but images suffer from the glare of headlights from oncoming vehicles.

As for the technology of choice in the future, the answer will be dictated by which is able to deliver sufficient performance for the function in mind, for example pedestrian detection. Until NIR systems are able to meet this goal, FIR systems will probably remain the preferred choice, despite being the more costly approach.

However, given that the capabilities of NIR and FIR partly complement one-another, the optimal system may ultimately involve a fusion of both technologies; for example, using an NIR system for generating an image, with a lower-resolution FIR sensor for purposes such as pedestrian detection. Initial impressions and favourable and suggest that both approaches score over their earlier counterparts but prices will need to fall dramatically if either system is to penetrate the high volume sectors of the car market.