ADAS and autonomous vehicles with distributed aperture radar

The automotive landscape is evolving, and vehicles are increasingly defined by advanced driver-assistance systems (ADAS) and autonomous driving technologies. Moreover, radar is becoming increasingly popular for ADAS applications, offering multiple benefits over rival technologies such as cameras and LiDAR.

It’s a lot more affordable, and it also operates more efficiently in challenging conditions, such as in the dark, when it’s raining or snowing, or even when sensors are covered in dirt. As such, radar sensors have become a workhorse for today’s ADAS features such as adaptive cruise control (ACC) and automatic emergency braking (AEB).

However, improved radar performance is still needed to ensure reliability, safety, and convenience of ADAS functions. For example, the ability to distinguish between objects like roadside infrastructure and stationary people or animals, or to detect lost cargo on the road, are essential to enable autonomous driving features. Radar sensors must provide sufficient resolution and accuracy to precisely detect and localize these objects at long range, allowing sufficient reaction time for a safe and reliable operation.

A radar’s performance is strongly influenced by its size. A bigger sensor has a larger radar aperture, which typically offers a higher angular resolution. This delivers multiple benefits and is essential for the precise detection and localization of objects in next-generation safety systems.

Radar solutions for vehicles are limited by size restrictions and mounting constraints, however. Bigger sensors are often difficult to integrate into vehicles, and the advent of electric vehicles has resulted in front grills increasingly being replaced with other design elements, creating new constraints for the all-important front radar.

With its modular approach, distributed aperture radar (DAR) can play a key role in navigating such design and integration challenges. DAR builds on traditional radar technology, combining multiple standard sensors to create a solution that’s greater than the sum of its parts in terms of performance.

Figure 1 DAR combines multiple standard sensors to create a more viable radar solution. Source: NXP

The challenges DAR is addressing

To understand DAR, it’s worth looking at the challenges the technology needs to overcome. Traditional medium-rage radar (MRR) sensors feature 12-16 virtual antenna channels. This technology has evolved into high-resolution radars, which provide enhanced performance by integrating far more channels onto a sensor, with the latest production-ready sensors featuring 192 virtual channels.

The next generation of high-resolution sensors might offer 256 virtual channels with innovative antenna designs and software algorithms for substantial performance gains. Alternative massive MIMO (M-MIMO) solutions are about to hit the market packing over 1,000 channels.

Simply integrating 1000s of channels is incredibly hardware-intensive and power-hungry. Each channel consumes power and requires more chip and board area, contributing to additional costs. As the number of channels increases, the sensor becomes more and more expensive, while at the same time, the aperture size remains limited by the physical realities of manufacturing and vehicle integration considerations. At the same time, the large size and power consumption of an M-MIMO radar make it difficult to integrate with the vehicle’s front bumper.

Combining multiple radars to increase performance

DAR combines two or three MRR sensors, operated coherently together to provide enhanced radar resolution. The use of two physically displaced sensors creates a large virtual aperture enabling enhanced azimuth resolution of 0.5 degrees or lower, which helps to separate objects which are closely spaced.

Figure 2 DAR enhances performance by integrating far more channels onto a sensor. Source: NXP

The image can be further improved using three sensors, enhancing elevation resolution to less than 1 degree. The higher-resolution radar helps the vehicle navigate complex driving scenarios while recognizing debris and other potential hazards on the road.

The signals from the sensors, based on an RFCMOS radar chip, are fused coherently to produce a significantly richer point cloud than has historically been practical. The fused signal is processed using a radar processor, which is specially developed to support distributed architectures.

Figure 3 Zendar is a software-driven DAR technology. Source: NXP

Zendar is a DAR technology, developing system software for deployment in automobiles. The performance improvement is software-driven, enabling automakers to leverage low-cost, standard radar sensors yet attain performance that’s comparable to or better than the top-of-the-line high-resolution radar counterparts.

How DAR compares to M-MIMO radars

M-MIMO is an alternative high-resolution radar solution that embraces the more traditional radar design paradigm, which is to use more hardware and more channels when building a radar system. M-MIMO radars feature between 1,000 and 2,000 channels, which is many multiples more than the current generation of high-resolution sensors. This helps to deliver increased point density, and the ability to sense data from concurrent sensor transmissions.

The resolution and accuracy performance of radar are limited by the aperture size of the sensor; however, M-MIMO radars with 1,500 channels have apertures that are comparable in size to high-resolution radar sensors with 192 channels. The aperture itself is limited by the sensor size, which is capped by manufacturing and packaging constraints, along with size and weight specifications.

As a result, even though M-MIMO solutions can offer more channels, DAR systems can outperform M-MIMO radars on angular resolution and accuracy performance because their aperture is not limited by sensor size. This offers significant additional integration flexibility for OEMs.

M-MIMO solutions are expensive because they use highly specialized and complex hardware to improve radar performance. The cost of M-MIMO systems and their inherently unscalable hardware-centric design make them impractical for everything but niche high-end vehicles.

Such solutions are also power-hungry due to significantly increased hardware channels and processing requirements, which drive expensive cooling measures to manage the thermal design of the radar, which in turn, creates additional design and integration challenges.

More efficient, cost-effective solution

DAR has the potential to revolutionize ADAS and autonomous driving accessibility by using simple, efficient, and considerably more affordable hardware that makes it easy for OEMs to scale ADAS functionality across vehicle ranges.

Coherent combining of distributed radar is the only radar design approach where aperture size is not constrained by hardware, enabling an angular resolution lower than 0.5 degrees at significantly lower power dissipation. This is simply not possible in a large single sensor with thousands of antennas, and it’s particularly relevant considering OEM challenges with the proliferation of electric vehicles and the evolution of car design.

DAR’s high resolution helps it to differentiate between roadside infrastructure, objects, and stationary people or animals. It provides a higher probability of detection for debris on the road, which is essential for avoiding accidents, and it’s capable of detecting cars up to 350-m away—a substantial increase in detection range compared to current-generation radar solutions.

Figure 4 DAR’s high resolution provides a higher probability of detection for debris on the road. Source: NXP

Leveraging the significant detection range extension enabled by an RFCMOS radar chip, DAR also provides the ability to separate two very low radar cross section (RCS) objects such as cyclists, beyond 240 m, while conventional solutions start to fail around 100 m.

Simpler two-sensor DAR solutions can be used to enable more effective ACC and AEB systems for mainstream vehicles, with safety improvements helping OEMs to pass increasingly stringent NCAP requirements.

Perhaps most importantly for OEMs, DAR is a particularly cost-effective solution. The component sensors benefit from economies of scale, and OEMs can achieve higher autonomy levels by simply adding another sensor to the system, rather than resorting to complex hardware such as LiDAR or high-channel-count radar.

Because the technology relies on existing sensors, it’s also much more mature. Current ADAS systems are not fully reliable—they can disengage suddenly or find themselves unable to handle driving situations that require high-resolution radar to safely understand, plan and respond. This means drivers should be on standby to react and take over the control of the vehicle suddenly. The improvements offered by DAR will enable ADAS systems to be more capable, more reliable, and demand less human intervention.

Changing the future of driving

DAR’s effectiveness and reliability will help carmakers deliver enhanced ADAS and autonomous driving solutions that are more reliable than current offerings. With DAR, carmakers will be able to develop driving automation that is both safer and provides more comfortable experiences for drivers and their passengers.

For a new technology, DAR is already particularly robust as it relies on the mainstream radar sensors which have already been used in millions of cars over the past few years. As for the future, ADAS using DAR will become more trusted in the market as these systems provide comprehensive and comfortable assisted driving experiences at more affordable prices.

Karthik Ramesh is marketing director at NXP Semiconductors.

Antonio Puglielli is VP of Engineering at Zendar.

Related Content

• Radar Basics: Range, Pulse, Frequency, and More
• Is Digital Radar the Answer to ADAS Interference?
• Cameras, Radars, LiDARs: Sensing the Road Ahead
• Challenges in designing automotive radar systems
• Automated Driving Is Transforming the Sensor and Computing Market
• Implementing digital processing for automotive radar using SoC FPGAs

The post ADAS and autonomous vehicles with distributed aperture radar appeared first on EDN.