FreightWaves

View Original

LiDAR wars update: Quanergy CEO takes on Velodyne

3D Point Cloud data generated by a Quanergy LiDAR sensor. (Image: Quanergy)

See this content in the original post

FreightWaves has covered the intense competition in the LiDAR industry since last year: we’ve reported on the settling of the Uber-Waymo lawsuit and new product releases by Velodyne. LiDAR--or 'Light Detection And Ranging'--is a radar-like technology that uses lasers to map the surroundings of autonomous vehicles.

One of the most interesting manufacturers in the space is Quanergy, a solid-state LiDAR hardware and software company based in Sunnyvale, California, with roots in the defense and aerospace industries. Quanergy announced the debut of its proprietary perception software platform, Qortex, at this year’s Consumer Electronics Show in Las Vegas. 

“Deciphering data from 3D point clouds is one of the most difficult aspects of using LiDAR technology,” said Jeroen Floor, vice president of software engineering at Quanergy. “As the central perception engine of Quanergy’s products, Qortex drastically improves how we understand these inputs by detecting and interpreting point clouds and providing actionable data from what our systems see.”

Quanergy says that their solid state LiDAR sensors—which have no moving parts, and use an optical phased  array laser—are cheaper, more reliable, and higher performing than any mechanical LiDAR systems on the market. FreightWaves spoke to Quanergy Founder and CEO Dr. Louay Eldada by phone.

“The big deal about solid state is, number one, delivering the performance needed in a primary sensor—it has the highest possible level of reliability and the lowest achievable cost. No moving parts. And with fully automated manufacturing, we can get the price down to $250 in volume,” said Eldada. 

With regard to Quanergy’s new software, Eldada explained his team’s deep pedigree in high-end defense and aerospace applications. Eldada’s doctoral work in optical engineering at Columbia University was on time-of-flight LiDAR navigation systems for missiles, which have to map complex topologies in three dimensions at very high speeds. Eldada said that DARPA realized that radio frequency guidance could be jammed, but semi-autonomous missiles mapping their own surroundings with lasers in real time were not vulnerable to the same countermeasures.

Check out this video of real-time mapping data of the Golden Gate Bridge generated by a Quanergy sensor in a single pass:

Eldada also criticized LiDAR software makers who do perception by flattening 3D data into 2D images. “People who did 2D computer vision for a living and got their PhDs in it cannot simply switch to 3D vision,” Eldada said. In reference to Velodyne’s customers, who keep asking for more layers, Eldada said, “The game should not be about number of layers. The focus on number of layers—16 to 32, to 64, to 128... Velodyne's customers in the robotics and automotive supplier segments are using video-based 2d perception software developed for cameras and applied it to 2D images generated by 3D sensors. They aren’t doing 3D perception on a 3D point cloud. It’s a brute force way of going down the same path that isn’t really that advanced.”

Finally, the Quanergy CEO asserted that mechanical LiDAR sensors—sensors that have rotating components to capture a 360 degree field of view—will never be commercially viable. “No automaker will use a mechanical LiDAR,” said Eldada. “They’re only using them in pilots and to develop software. You’ll not see a single model using a mechanical LiDAR above level 2 automation. Level 4 autonomy is what we’re all talking about when we talk about productivity gains while driving, mobility for disabled and blind people.” Quanergy’s partners include Mercedes-Benz, Renault Nissan, Hyundai, and Delphi. Meanwhile, Velodyne’s sensors are being used by Ford and Volvo, Mercedes-Benz, and even Radiohead, whose music video for “House of Cards” was filmed with Velodyne’s HDL-64E.

Eldada claimed that mechanical LiDAR sensors are unsuitable for commercial automotive applications because the mean-time-to-failure (MTF) on the rotating components is far too low for industry requirements. Eldada said that mechanical sensors have an MTF of between 1,000 to 3,000 hours of operation, while automakers want an MTF of at least 13,000 hours. Eldada said that Quanergy guarantees an MTF of 100,000 hours for its solid state LiDAR sensors. The range of Velodyne’s mechanical sensors, Eldada also said, were much lower than solid state sensors: “We have 3.5 cm spot size at 100 meters. For Velodyne, at 100 meters, you and your car are the same pixel. Velodyne only gives you an effect range of 80 to 120 meters—call it 100 meters. Quanergy offers 150-200 meters.”

We asked Velodyne’s leadership to respond to Eldada’s attacks on their product. Mike Jellen, Velodyne’s President and Chief Commercial Officer, wrote, “Velodyne LiDAR Inc. leads the industry in research, development and manufacturing of high-quality, real-time 3D LiDAR sensors. Velodyne operates from five facilities globally with over 500 employees and revenue in excess of $175M per year. Velodyne is the trusted supplier to the global autonomous vehicle / ADAS market, and has been for over ten years.” When asked to address the technical claims made by Eldada, Velodyne responded, “We have no further statements on this topic.”

Stay up-to-date with the latest commentary and insights on FreightTech and the impact to the markets by subscribing.

See this content in the original post