MAKING WAVES

Search

Metawave demonstrates first 77GHz analog beamsteering radar for autonomous vehicles

Read the full article on therobotreport.com By Eugene Demaitre | February 4, 2020

PALO ALTO, Calif. — Metawave Corp. today announced that it has successfully demonstrated the first analog beam-steering radar for automotive applications. SPEKTRA is designed to clearly detect vehicles beyond 300 meters and pedestrians beyond 200 meters with a high angular resolution.


Founded in 2017, Metawave has laboratories in Carlsbad and Palo Alto, Calif. The company has filed more than 150 patent applications to date and received its first patent in 2019. It raised funding in September 2017, May 2018, and November 2019, with past participation from Toyota AI Ventures and Denso Corp. The radar demonstration occurred at CES 2020.


SPEKTRA is the first analog beam-steering radar system capable of distinguishing between objects in difficult driving scenarios and in all weather conditions, making cars safer and smarter, said Metawave.


The company added that its TURBO active repeaters and ECHO passive reflectors can enable faster, more efficient 5G deployments, bringing connectivity to billions of users, indoors and out.

Metawave SPEKTRA uses precise steering


Metawave claimed that its MARCONI phase controller is the first of its kind and is at the heart of SPEKTRA. It was developed at 77GHz and is augmented by its proprietary adaptive-phase and amplitude-calibration systems for precise 0.1° steering angles.


“Most startups shine a beam, get a reflection, and have the processor do everything, such as separating objects and figuring out ranges,” explained Metawave founder and CEO Maha Achour. “That’s suboptimal, with limited detection of 100 to 120 meters for pedestrians and cars at 250 meters.”


SPEKTRA’s high angular resolution enables it to distinguish between objects that are very close together, said the company. Its narrow transmit and receive beams enable it to rapidly focus on all objects in the vehicle’s field of view (FoV) with high accuracy while avoiding interference in the pure analog domain.


“In the digital space, many radars can’t steer the beam more than 8°,” Achour told The Robot Report. “We can separate two lanes at 300 meters with a 1.2° resolution. No other sensors available on the market today with good power consumption and range can attain this performance.”


Only with focused beams and small cross-sections can a radar detect objects, such as pedestrians and hazardous road objects, at long ranges and in all weather conditions, said Metawave. Furthermore, the narrow beams and high angular precision allow SPEKTRA to track cross-traffic, a difficult problem for traditional radars.


“Metawave has essentially packaged a front-end radar using their own phase controllers, antenna array, and calibration control tables using a single transceiver radar chip to achieve impressive transmit and receive performance,” said Rich Dauber, director of 77GHz radar platforms at Veoneer, an automotive safety systems provider based in Sweden. “Their CES SPEKTRA demonstration showed longer radar performance with excellent angular accuracy and resolution.”

AWARE provides real-time object classification


Metawave said its platforms are integrated with AWARE, its artificial intelligence software for object classification on the road, and effective 5G network optimization.


“Analog beamforming was already used in 5G, but the radar space was using less-efficient methods than the telecom space,” Achour said. “We brought that knowhow and built a multimode radar, where range is always driven by scanning radar. 5G will be critical to automotive — maybe not immediately — but it helps with low-latency, high-bandwidth edge compute, relieving cars of the processing burden in high-density areas.”


“We started looking into object classification even before our steerable radar was developed,” recalled Achour. “With different snapshots, we can create much better signatures for an object, distinguishing a car from a truck or a motorcycle from a bicycle.”


In the coming months, Metawave plans to launch AWARE, giving SPEKTRA real-time object classification and labeling for more efficient and faster ADAS processing because of its proprietary machine learning sensor-fusion platform. Combined, SPEKTRA will be accurate in bad weather and at highway speeds, said the company.


SPEKTRA’s ability to detect and classify objects beyond 300 meters is due to its unique ability to rapidly direct extremely narrow beams across the vehicle’s FoV. Metawave said it is the only automotive radar supplier that delivers a completely calibrated phased array and front-end solution using a single conventional Frequency-Modulated Continuous Wave (FMCW) radar transceiver chip.


This allows for more concentrated beam-forming signals than conventional radar, said Metawave. Traditional radars use four or more transceiver chips for digital beamforming, or they “flash” a wide FoV, limiting their range and preventing their ability to accurately distinguish between objects.


Metawave PoC available


Metawave is working with select customers in its SPEKTRA proof-of-concept (PoC) evaluation program. This is intended to automotive leaders to test the next-generation analog radar for autonomous driving reliably and affordably, while having access to Metawave experts and its labs to bring the SPEKTRA radar to market.


By offering the flexibility to extract raw data at various receiver stages, customers can run their own algorithms and seamlessly process SPEKTRA data with their software stacks.


“Our solution is like a building block, and our hardware can provide data to customers like Waymo,” said Achour. “Most Tier 1 suppliers provide post-processed radar data, but that’s not sufficient. They should have a range of points to tap into the signal, and the software stack needs fine-grained information.”


“Seeing the excitement during our CES demonstrations reassured us that we followed the right path that no one else dared to tackle,” he said. “Once the market saw our beamforming and modular architecture, they understood it.”


“The next step is to demonstrate two-dimensional or horizontal and vertical steering, which will enable us to detect tires or separate a stalled vehicle on a bridge,” Achour said. “As we get to 2021, we want to integrate the antenna array with the transceiver chip as everything else is perfected.”



Why self-driving cars need radar


Beyond the debate over whether lidar or cameras are best for ensuring safety, radar is emerging as a critical sensor type for autonomous vehicles, according to Metawave. The global automotive radar market will reach $6.61 billion by 2021, predicts Markets and Markets.


“Radar has been in cars for decades. It’s a fundamental part of cruise-control systems,” said Achour. “The main requirement we hear over and over is range — trucks need a longer time to stop, as do cars on the freeway. Plus, lidar and cameras suffer from environmental conditions such as fog, rain, or smog.”


“All sensors will be required until sensor fusion gets smarter,” he added. “At Level 2++ autonomy, cost is a factor, but when you start getting to higher levels, a camera can never be rated as a fault-proof sensor. With the FMCW waveform, we can derive the velocity and angle of arrival, so the radar can be rated.”


“Metawave is making incredible strides in demonstrating what advanced radar is capable of achieving,” said Tony Cannestra, director of corporate ventures for the North American Research and Engineer Center at Denso. “Automakers and Tier 1 suppliers are looking for a new kind of sensor that can detect and classify at long ranges, separate stationary objects at long range, and perform well in inclement weather conditions and in the dark.”


“Metawave has proved analog beam-steering radar can help achieve these capabilities,” he added. “We look forward to continuing our work with them, and other startups, to make the world’s roads safer through new technology.”


SPEKTRA is designed to enable ADAS safety features, including left-turn assist, blind-spot monitoring, automatic emergency braking, adaptive cruise control, traffic-jam pilot, highway pilot, automatic emergency braking, lane assist, adaptive cruise control, and more. Radar will remain critical for reliable and low-cost sensors, with suitable form factors for commercial viability, said Metawave.


“Metawave continues to demonstrate innovation in analog beam-steering technology, and the 77GHz phase controller is an important component to enable this technique,” said Ritesh Tyagi, head of the Automotive Innovation Center at Infineon Technologies AG. “Infineon looks forward to working with new RFIC [radio-frequency integrated circuit] developments like these and use them in conjunction with its market-proven transceiver solutions to bring the technology to the market.”

metawave-red-white-shadow.png