Every year, 1.25 million people  are killed by car accidents, and human errors (drunk driving, speeding, ignoring traffic signals, and texting while driving) are responsible for more than 94 percent of these fatal accidents. This loss of 1.25 million lives per year is equivalent to seven, 500-passenger aircraft crashing every day.

To reduce the number of car accidents to as close to zero as possible, car makers, automotive suppliers, governments, academics, and even non-automotive technology providers are jointly developing advanced driver-assistance systems (ADAS) and ultimately autonomous vehicles.

This new automotive ecosystem is combining a wide variety of advanced technologies such as:

• Sensor fusions with radio detection and ranging (radar), light detection and ranging (lidar), and optical sensors (cameras).
• High speed information systems integrating automotive Ethernet networking, powerful signal processing, high definition (HD) mapping with high precision navigation, and artificial intelligence (AI).
• Communications for vehicle-to-vehicle (V2V), vehicle-to-network (V2N), vehicle-to-infrastructure (V2I), vehicle-to-pedestrian (V2P), vehicle-to-utility (V2U), and eventually vehicle-to-everything (V2X).

Each sensing technology has benefits and limitations, and the automotive industry can’t depend on just one sensing technology. Most leading players of the autonomous driving industry combine all three sensing technologies to make sure their autonomous driving systems get a mix of reliable data across range, resolution, and robustness.

Autonomous vehicles will bring enormous benefits by saving lives and making driving safer, but developers need to prove that their mission-critical technologies are reliable and safe. Concerns were raised after a self-driving car was recently involved in the first fatal accident with a pedestrian. Citizens and regulators will constrain the mass-deployment of autonomous vehicles without evidence that machine errors are close to zero, no matter how much they can mitigate human-driven accidents. To address these concerns, designers and engineers must implement the most reliable, all-conditions sensor technology. They must also validate and demonstrate accuracy and dependability by using the best simulation and test solutions.

Three Major Sensing Technologies

Among all sensing technologies, lidar is the newest and is generating excitement in the autonomous vehicle market. It provides the most precise 3D mapping using laser light and can scan a 360-degree space around the autonomous car up to a 100-m range (300 ft). Some lidar systems provide as many as 64 channels and over a million points of scans per second. This amount of information provides high accuracy of 2 cm (~ 1 in.) for reacting to a changing environment.

However, lidar-based sensing has not been proven yet in mass-market applications and relies on moving parts for rotating 360-degree scans. Lidar generates a massive volume of data, and tremendous signal processing power and data management sub-systems are required. These sensors are also still very expensive, despite recent announcements of more economic versions of lidar.

Optical camera sensing is the lowest cost technology, although image processing cost can be expensive. It provides identification and classification of objects and reads traffic signs. However, camera-based sensing technology is affected by weather and other environmental conditions, and therefore cannot be relied on at all times. Its limitation was made clear by a tragic accident involving a camera-based, semi-autonomous driving vehicle that failed to discern a white truck on the road in front of a white sky background.

Automotive radar is the most reliable technology at detecting objects’ distance (range) and motion, including velocity and angle in almost every condition. It uses reflected radio waves to detect obstacles behind other obstacles and has low signal processing requirements. The radar technology has been proven over decades in many safety-conscious industries such as aviation, air traffic control, maritime transport, law enforcement, and of course automotive.

However, traditional 24 GHz narrow-band automotive radar has limited capability to differentiate objects and separate humans, dogs, other cars, or even mailboxes. While radar-only sensing technology does not provide enough data to enable a full autonomous driving system, automotive radar has been delivering benefits in ADAS by mitigating human driver errors, making it a strong bridging platform before fully autonomous vehicles become mainstream. Current mass-market applications of ADAS by automotive radar sensing technology include auto-emergency braking systems, forward collision warning, blind spot detection, lane change assist, rear collision warning systems, adaptive high-speed cruise control on highways, and stop and go cruise control in bumper-to-bumper traffic.

	Radar	Camera	Lidar
Technology	Detection—distance (range) and motion (velocity and angle) by radio waveforms	Recognition, classification by images	360-degree 3D view by laser light
Application	Adaptive Cruise Control, Automatic Emergency Braking Systems, Blind Spot Detection, Parking Assistance	Traffic Sign Recognition, Land Keep Systems, Parking Assistance, Blind Spot Detection, ACC, AEBS	Emergency Brake Assist for Pedestrian, Crash Imminent Braking, Mapping
Advantages	Proven over decades Reliable in all environmental conditions Small and light-weight Longer detection range	Lower cost Mass-market technology Smaller sensor size High resolution Color recognition Imaging processing	High accuracy High resolution Intelligent signal processing with large amount of captured data
Limitations	Limited information on detected objects (shape and movement only) Lower resolution than lidar and optical	Performance varies with environment (e.g. rain, snow, lighting)	Unproven in massive deployments Expensive and large sensor Expensive and complex signal/data management Affected by weather (e.g. rain, snow, lighting)

Table 1: Comparison of wave-based radar, image-based camera, and light-based lidar sensing technologies.

Although every technology plays an important role in the development of autonomous vehicles, sensors are critical to achieving their goal to improve road safety. Out of the all sensors available, the automotive radar uniquely determines all three target variables of velocity, range, and angle, regardless of light, dark, sunny, or rainy conditions.

Wider Bandwidth Automotive Radar
Automotive radar sensing technology, mainstreamed by the 24 GHz narrow band sensors, is now rapidly evolving toward high frequency 76-81 GHz band and wide 5 GHz bandwidth, millimeter wave, Frequency Modulated Continuous Waveform (FMCW), and beamforming antenna.

While the 76 GHz is used for long-range detection, the 77-81 GHz band is used for short-range, higher-precision detection. It is important to understand the magnitude of improvement delivered by the higher frequency, wider bandwidth, advanced automotive radar systems.

The error in distance measurement and minimum resolvable distance are inversely proportional to the bandwidth. Transitioning from 24 GHz to 79 GHz delivers 20x better performance in range resolution and accuracy. For instance, the improvement in range resolution from the 75 cm of a 24 GHz system to the 4 cm of a 79 GHz system offers better detection of multiple objects that are close together (Figure 1).

Also with smaller wavelength, the resolution and accuracy of velocity measurement increases proportionally. Therefore, by transitioning from 24 GHz to 79 GHz, velocity measurements can be improved by a factor of 3x.

Another advantage of the transition from legacy 24 GHz to 79 GHz systems is the gain in size and weight. With the wavelength of 79 GHz signals being a third of a 24 GHz system, the total area of a 79 GHz antenna is one-ninth of a similar 24 GHz antenna. Developers can use smaller and lighter sensors and hide them more easily for better fuel economy and car designs.

Test comparisons between 1 GHz and 4 GHz bandwidths (Figure 2) clearly show that only the higher bandwidth solution can measure two different objects, as close as 10 cm apart. The lower bandwidth radar is not able to detect two different objects and provides incorrect data to the driver or autonomous driving system, leading to the wrong decision. If a man and his dog are walking closely together and the dog suddenly jumps onto the road, only the wider bandwidth radar (test on right) can detect both separately and provide the correct information to the driver or the autonomous driving system. On the other hand, the narrower bandwidth radar (test on left) provides wrong or confusing information, possibly leading to a tragic accident.

With higher resolution and range, automotive radar leads to safer driving in any condition. A driver sending a text while on a highway at 55 mph can be oblivious of the traffic long enough to pass an entire football field. While the distracted driver may fail to notice a traffic slowdown, the radar-based ADAS system will alert with acoustic and visual warnings before a potential crash. If the driver doesn’t take any action, the ADAS system can initiate a full autonomous braking response to avoid the collision or reduce the force of impact. Only a millimeter wave wide bandwidth advanced automotive radar can perform this critical mission day and night, in sunny or rainy weather, for short- and long-range object detection.

Radar the Clear Solution

Automotive radar is today’s leading sensing technology for increasing driving safety in all environmental conditions. Advanced automotive radar takes advantage of 76-81 GHz band higher frequency and 5 GHz wider bandwidth for better resolutions, smaller and lighter sensor design, applicable to current and future ADAS, and upcoming autonomous driving systems.

To bring about an ideal world of zero-fatal car accidents, automotive radar developers need to optimize their design and test methods, and prove the reliability and safety of their solutions.