The ADAS Sensor Conference 2019 provided numerous insights into the current state of the industry and technology from both financial and technical perspectives. The combination of sensors continues to be the trend with most still favoring the combination of radar, image and lidar. However thermal was mentioned especially for applications that differentiate humans and animals from objects. Far infrared (microbolometers) sensors got some attention for expanding their applications beyond traditional night vision. Beyond radar, image and lidar sensors, high precision IMUs and GNSS can be added to the sensor suite and allow for more accurate placement of the car on HD maps. However wide scale deployment of high precision GNSS need to involve strategies to improve localization and convergence time. Analog Devices presented the interesting possibility of using ground penetrating radar to look under the vehicle at infrastructure as another means to locate its position with increase precision.
Lidar was a very common sensor technology discussed and the consensus is that lidar will continue to evolve from mechanical systems through MEMs based mirror systems to solid state. Regarding solid state beam steering is the area receiving the most attention largely because it is the most difficult part of the sensor. Micralyne discussed their Spatial Light Valves which have an advantage of increasing a lidar sensor’s resonant frequency above that of the vibrations typical in automotive applications. Another interesting point that came up regarding long range lidar is that larger apertures reduce beam divergence. Obvious longer-range applications require better angular accuracy.
Woodside Capital addressed the financial side of the industry and estimated that are at least 70 lidar suppliers currently and that number of suppliers and competing technologies is unsustainable. An example of that is the capital available to support lidar companies is tightening because AI is competing for the same capital investment. Therefore, it will be difficult for this large number of lidar companies to continue to get funding for development which will result in consolidation and contraction.
Magneti Marelli talked about V2X as complementary the ADAS sensor suite. GM is currently in production with DSRC and Toyota and VW have also released plans to support DSRC. GM is implementing DSRC in a standalone ECU but over time expect it to be integrated into a suitably rated ASIL module for example an ADAS ECU. C-V2X is also capable of communications with the infrastructure and other vehicles and does not require a SIM card or a data plan. Ford seems to be favoring C-V2X for development. The sensitivity and importance of cybersecurity is increasing; Green Hills has created an encryption key for V2V communications.
In addition to the theoretical issues of how to build a better lidar sensor or what is needed to more precisely place a vehicle on a HD map, City Auto Glass presented the very practical issues of recalibrating sensors in the field. Recalibrating sensors after a car is in an accident seems sensible, but what about the need to recalibrate the sensor mounted behind the rear-view mirror after a stone cracks a windshield requiring it to be replaced? Regarding the sensors mounted on the outside of the car they may require recalibrations after the suspension is adjusted.
Another practical topic mentioned through the conference was how to keep sensors clear of water and snow. While it’s accepted that radar sensor can see through falling snow, they cannot see through snow that has built up on the sensor. One idea is to coat the sensor with a material to which water does not stick. If the sensor sheds water than snow, ice, mud should slide off the sensor keeping it clear.