The DVN-Lidar Deep Dive III was held in San Francisco on 30 Aug, with a strong focus on the FMCW technology.
With 26 participants registered, we had a large group of experts from automakers (Ford); lidar suppliers (Valeo, Aeva, Opsys, Moblieye, Koito, Red Creamery); chip makers. All had the opportunity to discuss in small groups, before sharing arguments in the plenary meeting.
Here we bring you a summary of the presentations:
Valeo / Waqas Malik
“Making Autonomous Driving a Reality”
- Lidar enables a wider ODD than traditional L2 systems
- Point cloud density is a key feature for high performance lidars.
- The 3rd generation of Scala achieves 12.5 million points/s
- The extension of the ODD is the next challenge for L3 systems including vehicle speed and adverse weather conditions
- Extending the ODD requires multiple sensor technologies (camera, radar, lidar), and redundancy to achieve the safety goal.
Virginia Tech Transportation Institute / Matthew Palmer
“Advanced Research Project about Lidar Congestion”
- The research project is funded by NTHSA, conclusions are expected end of 2024
- The goal is to examine the physics behind Lidar congestion interference and the state of understanding of the automotive industry of lidar congestion, interference, system impact, and mitigation strategies
Siemens / Tom Witsenboer
“Physics Based Raw sensor Simulation”
- Simulation is necessary to develop safe functions for AVs, to validate all the scenarios, including the edge cases that you cannot test with a real car
- Lidar Simulation requires a proper sensor model from the supplier and a good simulation of the environment (a database is necessary to store the reflectivity of the materials)
- Simcenter Prescan enables modelling of the entire pipeline, dynamics of the environment, sensor outputs, technology specific sensor effects, and material properties.
Hamamatsu / Slawomir Piatek
“FMCW – a closer look at the principles of operation and challenges”
- Introduction to the FMCW concept
- Current Automotive Lidar technologies
- Closer look at FMCW
- Examples of Photonic integration
- Expectations for FMCA Lidars
SILC / Ralf Munster
“Enabling Mass Market of FMCW Lidar through Chip Integration”
- FMCW allows a direct detection of moving objects
- FMCW has a longer range than TOF Lidars
- FMCW is not impacted by the sun, has a very low interference risk, and has improved performance in bad weather conditions
- Photonics integration and a SoC approach allows to be cost competitive despite a higher complexity than TOF
Indie semi / Setu Mohta
“Purpose-Build Signal SoC Platforms to Catalyse the Deployment of Coherent Lidars”
- Indie is an automotive fabless semiconductor company supporting all kind of sensor technologies (cameras, radars, lidars)
- FMCW requires 4 to 16 ADC converters and a Purpose-build Silicon that understands Lidar natively and can scale
- Indie SoC integrates Lidar ADC/DACs for Lidar Tx/Rx, 4x Hardware DSP (FFTs), 4 software DSPs, a 32-bits MCU (Arm), CAN-FD, Ethernet
- Indie SoC allows the deployment of competitive FMCW Lidars
Conclusions: the technical discussions between multiple experts and suppliers allowed to have a clear view on the FMCW technology, the pros and cons, and its maturity level. A-Samples are ready at SILC which means the technology could be in 2027-2028 on the market. Test results are required to validate the pros and cons of the technology (benefit of a direct detection of moving objects, but a somewhat lower point density)