Putting it into practice - 100 million times, ON Semiconductor's sensors protect road safety

As the world's population grows and more vehicles are on the road, improvements in traffic safety cannot come fast enough. Although the fatality rate has been reduced by more than half since 2000, from 135 to 64 per 100,000 vehicles, the total number of fatalities continues to climb. Today, 94% of accidents are attributed to driver behavior.

Today’s cars are safer than ever before. Air bags, mandatory seat belts, and improvements in vehicle structure and functional design make it more likely that drivers and passengers will walk away from a crash.

Better braking and steering subsystems, as well as common innovations such as anti-lock braking systems (ABS) or electronic stability control (ESC), rely on accurate sensors to improve safety, while advanced driver assistance systems (ADAS) make traffic accidents less likely.

As cars become more autonomous, the aim is to reduce this risk further. The ultimate goal is fully autonomous driving, or “Level 5,” which effectively eliminates human error.

Vision Zero is a multi-national initiative that envisions no road crashes resulting in death or serious injury.

One of the main factors in improving automobile safety is the improvement of the level of electronic devices in the car. For example, there are currently more than 230 ON Semiconductor devices in each car on average worldwide.

If we are to achieve Vision Zero, we cannot be complacent and there is still much work to do. Although cars are safer, at least in developed countries, road traffic injuries remain the most common cause of death among people aged 5 to 29, with the World Health Organization (WHO) reporting that road traffic accidents kill 1.35 million people each year.

Sensor Integration

Twenty years ago, any sensor in your car was pretty simple. You had a fuel gauge that measured the level of fluid in the tank, and an engine temperature gauge. Along with your speedometer and some warning lights, maybe a tachometer, and that was probably it.

Today, many electronic sensors help keep you safe. For example, cameras and imaging sensors have many uses, including ADAS, rear view for safe backing and parking, and in-cabin monitoring.

This means that it is important for automakers, OEMs, and Tier-1s to work with suppliers that have a broad range of products so that they can select the best sensor for each application. The sensor should also be designed for critical missions and be able to operate over an extended temperature range.

Blind-Spot Detection: Blind Spot Detection

Backup Camera: Rear view camera

Car DVR: Driving Recorder

Driver Monitoring: Driver Monitoring

Lane Departure Warning: Lane departure warning

Pedestrian/Object Detection: Pedestrian/Object Detection

Collision Mitigation

Adaptive Cruise Control: Adaptive Cruise Control

Smart Headlight/Mirror: Smart headlight/mirror

Night Vision:

Figure 1: Image sensor applications in automobiles

Performance is also important—the sensor must have high enough resolution to capture enough detail for ADAS and other systems, and it must provide excellent image quality to cope with darkness, bad weather, glare and other issues.

For example, superior dynamic range can dramatically change the image sent from the sensor to the ADAS system processor (see Figure 2). It’s no exaggeration to say that it’s the difference between life and death if it means the car can more quickly identify a problem ahead.

Competition

ON Semiconductor: ON Semiconductor

Figure 2: Dynamic Range Comparison - Note the lack of detail at the end of the tunnel in the image on the left

In addition to image sensors , radar and LiDAR are essential tools in today's cars. Radar can be used for short-, medium- and long-range applications such as evasive steering, junction assistance and adaptive cruise control, which can look up to 250 m ahead. For each use case, choosing the right radar transceiver will ensure optimal performance.

LiDAR complements radar, with its photon detectors capable of generating images as well as 3D maps based on measuring time of flight (ToF). This enables LiDAR to provide high-resolution depth data, enabling object detection capabilities that are not possible using radar or cameras alone.

In practice, the best approach is often to combine multiple perception modalities in one vehicle: imaging, radar and LiDAR, and ultrasonic perception. Using multiple types of sensors , their strengths can work together, and redundancy is built in.

Put into practice - 100 million times

Let’s look at an example system: SUBARU’s EyeSight driver assistance system, which uses ON Semiconductor’s 1.2-megapixel AR0132AT CMOS image sensor.

The EyeSight system was first installed in SUBARU's Levorg model in 2014 and was subsequently offered in the Legacy, Forester, Impreza and SUBARU XV models.

EyeSight uses image sensors in its stereo camera system to enable safety features including adaptive cruise control, lane keeping assist and sway warning, pre-collision braking and pre-collision throttle management.

The system has won numerous awards, including the Japan New Car Assessment Program (JNCAP)'s Advanced Safety Vehicle Triple Plus (ASV+++) top rating.

ADAS is now a mainstream technology, not just for protection in high-end cars. This is evidenced by the fact that ON Semiconductor has shipped more than 100 million AR0132AT image sensors for driver assistance applications, including EyeSight.

No other vendor appears to have reached these numbers (and in 2018, ON Semiconductor had 81% market share in perception cameras for driver assistance), which shows the scale of technology adoption.

Drowsiness while driving is another leading cause of traffic accidents. Technology can help monitor a driver's performance and issue alerts or warnings if they appear to be driving erratically.

Another option is to use a camera-based system to observe the driver and trigger an alert if it detects signs of fatigue, such as closed eyes or drooped head. For example, a recent demonstration system integrated multiple image sensors, including ON Semiconductor's AR0144AT 1-megapixel sensor, to provide images to an in-vehicle system running artificial intelligence (AI) software.

The future of safe driving

Today, 28% of traffic accidents in the U.S. alone could be prevented with ADAS, and ON Semiconductor sensors already save more than 81,000 lives each year. That’s good, but it has to be better, and we’ll continue to work hard to improve sensors and collaborate with partners to make cars and roads safer.

Cost is also important; traffic accident rates are much higher in low-income countries, so any safety innovation should be widely affordable.

Going forward, we must also ensure that safety systems not only protect drivers and passengers, but also help reduce casualties among pedestrians, motorcyclists and cyclists.

Regulations and standards are recognizing this, such as the European New Car Assessment Programme (EuroNCAP 2020), and sensor technology can play an important role in protecting vulnerable road users.

The industry's long-term goal must be to work towards zero fatalities, and even zero injuries and zero incidents. There is still some way to go, but we are working hard to get there as quickly as possible.