Already established as a fundamental technology for advanced in-vehicle safety systems, mmWave radar is about to extend its reach into the transportation infrastructure.
The force behind Tesla, Space X, and the Boring Company is well known for exaggeration, but his projections also tend to become true–just not when he says they will. Space X is leading the commercial space industry, Tesla just made its first profit, and the Boring Company just let a few people experience travel is his first tunnel, So, betting against his goals is risky in the long term, which is when self-driving cars will actually proliferate, as enormous hurdles remain, equal to or more likely even greater than those the industry has yet experienced. And somewhere in the top 10 is traffic monitoring and management, a task in which millimeter-wave radar will a major role.
It was not long ago that very few people in any industry paid any attention to millimeter-wave frequencies, which are more than 10 times higher than those used for all types of wireless communications. The reasons for this disinterest are primarily the way electromagnetic energy (that is, radio signals) travels in space at these frequencies and the difficulty and cost and complexity of the semiconductors and other components required to operate there.
Consequently, only satellite-based remote sensors, some defense systems, and a small number of scientific applications dared venture into this vacant space. The only commercial terrestrial application has been the radars integrated into vehicles, a market that has grown exponentially in the last decade, and certain to continue as more radar sensors are used in every vehicle in the future.
However, some propagation characteristics of millimeter-wave signals that are problematic for communications are not as important when reliable transmission is required only over short distances, as it is for radar sensors operating at frequencies between 76 and 81 GHz used in vehicles and soon for traffic monitoring at management. Radar sensing can also be performed in low-light conditions and through smoke, fog, and rain, unlike LIDAR and cameras.
(This animation demonstrates how TI single-chip millimeter-wave (mmWave) technology can be used for multivehicle tracking across multiple lanes.)
Radar also provides greater resolution (5 cm or less), which translates into more accurate measurement of distance, velocity, and direction, and its detection range has been demonstrated up to 150 m. Detection and identification can be performed in a few hundred milliseconds or less, depending on the performance of the associated digital signal processing capabilities within the radar or an external subsystem.
Of course, the effectiveness of any type of sensor varies with the severity of environmental conditions, and these effects are often very difficult to measure. For example, rainfall attenuates signals at millimeter-wave frequencies at a rate that increases with frequency (Figure 1), precipitation rate, raindrop size, and the effect of ice, rain, and snow on the antenna. The latter has been an issue with vehicle radars as ice buildup is a common occurrence.
Rainfall must also be studied both generally and in concert with a specific location, as rainfall rates in some places can be very high at times, typically briefly, which obviously increases attenuation. Even so, attenuation over the distances required by traffic management systems is measured in feet rather than kilometers, and attenuation through scattering is only a few decibels, which is sufficiently low to make the 77 GHz region acceptable for intelligent transportation applications.
The use of millimeter-wave frequencies for radar has other benefits as well, such as the ability to penetrate objects, something no other technology can accomplish, and can distinguish between different types of objects in a given place, even if they are traveling at the same speed and are next to each other. The former capability is why millimeter-wave scanners are increasingly used by the TSA at check-in lines, where they can identify suspect objects, regardless of what materials they are made of. Millimeter-wave radars are also used to find objects buried at short depths in the ground and to detect activity behind walls made of common construction materials.
Radar and Traffic Management
Although a variety of techniques have been used over the years for traffic monitoring and vehicle detection, there have been few true advances. For example, inductive loop sensors (Figure 1) have long been used for vehicle detection at intersections. These systems consist of a coil of wire embedded in the road surface that is connected to a control system. electrical energy is passed through the coil at a frequency between 10 and 200 kHz and the presence of a vehicle, either before or as it reaches the intersection, is detected by a change in inductance, which trips a relay that typically changes the state of a traffic light from green to amber.
This is obviously a pretty crude technique and has drawbacks that make it too unreliable to be as a mission-critical sensor. For example, as every motorcycle rider knows, these sensors fail to detect their presence as the motorcycle is too small to sufficiency change the inductance. The same scenario applies to bicycles, scooters, and of course pedestrians as inductive loops can only detect metallic objects. In addition, as they are embedded in the road, maintenance or replacement requires digging them up (more common today than ever as the nation’s roads deteriorate), and a loop is required in every lane. The decision tree for these systems is also rudimentary as they cannot take into consideration any scenario other than the few they’re designed for.
More recently, video cameras have been employed for traffic monitoring and are an integral part of traffic management and autonomous vehicle systems in the future. When combined with high-performance signal processing they can provide extraordinary capabilities for detecting and analyzing everything from basic shapes to identifiable people. They have become a core element of surveillance systems throughout the world. However, they have the disadvantages of all vision-based systems such as degraded performance in poor weather, dependence on lighting conditions including shadows, and decreasing resolution with range.
In contrast, millimeter-wave radar has none of these limitations, is easy to install, far more accurate, can make decisions based on a huge number of potential situations that are limited only by signal processing capability and computation power. As they are mounted above rather than in the road, they can scan a wide area and potentially detect and analyze activities that may or may not become a problem depending on the actions of vehicles, pedestrians, and any other moving object. Interestingly enough, microwave (not millimeter-wave) radars were used nearly 50 years ago in some cities and their benefits were welcome. Why they were never widely deployed is a bit of mystery.
“Almost all cars produced will be autonomous in 10 years,” he said. “It will be rare to find one that is not in 10 years.” — Elon Musk
Radar sensors for traffic management fall into two general categories: pulsed Doppler and frequency-modulated continuous wave (FMCW). Within these broad types, radar manufacturers and universities have also developed their own coding schemes for radar-based traffic detection systems. Pulse radar sends short pulses and determines distance by measuring the time delay between transmitted and returned signal. FMCW radar constantly sends out linearly modulated signal and determines the distance based on the difference in transmitted and received frequency.
Most millimeter-wave radars for in-vehicle and traffic monitoring applications use FMCW radars because they require much lower RF output power and can “listen” while transmitting. They are or can be less expensive than pulsed Doppler radars and they have the benefit of better detection capability over the short distances required in automotive applications. Their limitations when measuring moving targets are overcome using coding schemes.
One of the benefits inherent in using millimeter-wave frequencies is their short wavelengths (just 3.8 mm at 77 GHz) that allow entire systems to be integrated within a single SoC rather than in multiple devices (Figure 2). Some of these devices are made with silicon germanium (SiGe), but the latest are based on either RF CMOS or BiCMOS.
This is a huge benefit for several reasons, not the least of which is cost, which has declined dramatically. Like all high-frequency circuits, integration within a single device reduces latency, susceptibility to interference, while reducing the bill of materials and the footprint of the complete product. As the radar and its antenna are very small, placement can be virtually anywhere, typically (but not necessarily) on a utility pole. Radar data can be communicated using a wired solution if available or wireless means such as Wi-Fi, and potentially cellular networks.
A good example of the SoC approach is the IWR1642 from Texas Instruments (Figure 3) that integrates broad functionality within a 10.4 x 10.4 mm flip-chip BGA package. The FMCW sensor, based on RF CMOS, operates from 76 to 81 GHz with up to 4 GHz of continuous chirp over two transmit and four receive channels, and delivers an RF output of 12.5 dBm. An ARM Cortex-R4F radio control system provides front-end configuration, control, and calibration, and a TI C674x DSP performs processing of the advanced algorithms that provide real-time detection, tracking and object classification. It can form the basis of a wide field of view (±50-deg.), 70-m range system that can detect small cars across four lanes and track their position and velocity as they approach the intersection and the stop bar.
The use of millimeter-wave frequencies has always been considered “just around the corner”, which indeed it was for decades. So, it’s rather remarkable not just that they have finally turned that corner but that commercial single-chip millimeter-wave radars are virtually mainstream systems today. They have become the fundamental sensors for advanced in-vehicle safety systems and are about to extend its reach into the external sensing applications that are equally vital for full vehicle autonomy in intelligent transportation infrastructure. They are even replacing less-advanced but traditionally less expensive sensors such as those based on ultrasound even though they are far more complex.