Koito and Quanergy collaborate to design automotive headlight concept with built–in solid-state LiDAR
07 January 2017
Koito Manufacturing Co., Ltd., the largest global maker of automotive headlights, and Quanergy Systems, Inc., a leading provider of LiDAR sensors and smart sensing solutions, are collaborating to design an automotive headlight concept with built-in Quanergy S3 solid state LiDAR sensors (earlier post). The Koito headlight with built-in sensors is on display at CES 2017.
The Koito headlights, which will be located on the corners of a vehicle, each incorporates two compact Quanergy S3 solid state LiDARs that perform sensing forward and to the side, and provide real-time long-range 3D views of the environment around the vehicle and the ability to recognize and track objects.
The sensors are seamlessly embedded inside the headlights and do not protrude or alter the exterior look of the headlights or the vehicle. Additionally, the headlight protects the sensors from dust, dirt and water, and headlight lens washers can be used to help ensure an unobstructed view for the sensors.
The LiDARs embedded in the headlights have the function of primary sensors in advanced driver assistance systems (ADAS) and autonomous vehicle (AV) systems. As primary sensors, the LiDARs are used for (a) perception, (b) mapping and localization, and (c) occupancy grid detection, path planning and navigation.
We are excited to be working with Koito on this groundbreaking headlight. We believe that headlights with integrated LiDAR sensors will help accelerate the commercialization of cars with autonomous driving capability and will help reinvent the driving experience.—Dr. Louay Eldada, Quanergy CEO
Quanergy’s solid state LiDAR uses an optical phased array as a transmitter, which can steer pulses of light by shifting the phase of a laser pulse as it’s projected through the array.
The S3 offers a number of capabilities that are software-controlled in real time:
- Adjustable window within total available field of view;
- Arbitrary distribution of points in point cloud; point density within a frame not necessarily uniform (e.g., denser distribution around horizon in vehicle);
- Random access for maximum SNR at receiver;
- Largest VFOV (matches 120 HFOV);
- Zoom in & out for coarse & fine view;
- Adjustable frame rate based on situation analysis; and
- Directional range enhancement based on location in pre-existing map (e.g., maximum forward range on highway, maximum sideways range at intersection).
The Quanergy S3 is the world’s first and only compact, low-cost, automotive-grade solid state LiDAR sensor with high reliability and superior capability. The S3 LiDAR sensor was recognized as the 2017 CES Best of Innovation Award grand winner in the vehicle intelligence category. The company’s sensors enable autonomous driving and perform real-time 3D mapping and object detection, classification and tracking.
I never much fancied the big revolving thingies on the roof, and Volvo and now Hyundai have them in the bumper, but this could give a much wider field of view.
Posted by: Davemart | 07 January 2017 at 06:37 AM
I'd put the as high as possible without sticking up like horns.
The higher they are the better a view they will have.
On the other hand, if you make headlights, you have to put them there.
A small bump on the roof, or just inside the car (and swept by the wipers) would do nicely.
Posted by: mahonj | 07 January 2017 at 10:45 AM
Very good idea!
Posted by: SJC | 07 January 2017 at 01:26 PM
Forward looking cameras and LIDARs in particular will gather dust and snow and subsequently malfunction unless they are placed behind lens washers like in the headlights or behind the front windshield washers. Sonar and radar will function despite dust and snow. I think Tesla’s solution with 3 cameras just below the roof behind the front windshield is ideal.
For real world autonomous cars the sensors need to be out of sight and they need to have a system for removing dirt and snow. All of Tesla’s cameras have a heater that removes snow and ice. I am not sure the side and back looking cameras in Tesla cars has a system for removing dust. However, they are oriented so that they are shielded from the wind when the car moves forward. Software could sound an alarm when it is time for the vehicle owner to manually clean these sensors to secure proper functioning.
Ford has made a solution that washes the camera see https://www.youtube.com/watch?v=wsikAvX91Ss
If needed, Tesla may get that system also for its side and backward looking cameras. Hopefully it is not needed. It adds cost and stuff that can break. Hopefully a monthly manual cleaning will do.
Posted by: Account Deleted | 08 January 2017 at 05:01 AM
Thin sensor bars could be incorporated into front and rear roofs to get all location data required. Self cleaning heated glass could be used to reduce the accumulation of snow, ice, dirt and dust.
A second set of censor could be incorporated in front and back lights?
Future ADVs will need many redundant sensors for accurate 360 degrees data.
Side censors would be required to warm drivers crossing red lights and/or not making stop.
Posted by: HarveyD | 08 January 2017 at 09:28 AM
You do not need high res 3d 360 degree imaging.
You need front arc (say +-45 30 degrees 3d+ hi res) + you need peripheral vision at lower res for changing lanes and approaching junctions.
Everything needs to work in rain and dust so most sensors need to be behind wipers. You will also need HDR so you can operate as you enter and leave tunnels etc.
Simple cameras are very cheap now so you could have an array of them across the top of the window and a few looking right and left and behind. Lidar is still (quite) expensive so you won't have too much of that. Processing power is cheap once you put it in an ASIC or as part of a vision chip. The big cost (IMO) will be programming the systems and adapting them to different driving styles and environments.
One reason that the big 3 are slower to bring out AVs is that they are in snowy parts of the world (Michigan) while Palo Alto is mostly snow free and easy to think about AVs.
Programming a driving style for cities where people break the traffic rules all the time will be a challenge (Delhi, Naples, Dublin anybody ...)
Geneva and San Jose might be a piece of cake compared to these (and many other) places.
Posted by: mahonj | 08 January 2017 at 03:18 PM
The camera censors may be cheap but the installation of them with data and power cabels and redundancy is not cheap. There are several miles of cables in a modern car and even more in a self-driving Tesla. Tesla charges 8000 USD for the self driving ability option and I think this is a price they also need to charge for Model 3 that will have an identical sensor/computer package. I expect the price will not go down in the next 5 years for the autonomous option because Tesla will want to have more computing power and more sensors in the next few versions of their self driving hardware. The current version of Tesla’s self driving package will probably always be limited to 95% of all roads and only be twice as safe as human driving. The goal is 100% of all roads on the planet and 10 times safer. We need better hardware for that.
Posted by: Account Deleted | 08 January 2017 at 11:37 PM
Yes CHANGE, the goal is 100% of all roads in all weather conditions and 10 times safer. I also strongly believe that much better hardware and software are required for that.
Tests can start in 2017/2018+ but a complet safe all weather system will probably not be around much before 2025 or so.
Posted by: HarveyD | 10 January 2017 at 03:42 PM
Most of the data communication (with/beteween units in ADVs) could be wireless via updated secured WiFi and/or Blue Tooth limited/close range technologies.
Many of the existing cables could be eliminated.
Posted by: HarveyD | 16 January 2017 at 05:46 PM