Tech companies are committed to produce affordable LiDAR sensors, reports Andrew Williams [Auto.Williams.2015.12.02]
In the ongoing quest to develop autonomous vehicles, LiDAR technology has emerged as a key enabling technology.
A number of companies are currently developing sophisticated LiDAR technology for use in autonomous vehicles, as well as in advanced driver assistance systems (ADAS).
Naturally, there are several providers fighting for recognition in this space and one such is Quanergy, which has recently launched its S3 LiDAR, the first integrated solid-state LiDAR system, claiming to bring “versatile, comprehensive, and affordable sensing” to autonomous cars. Another company heavily involved at the forefront of LiDAR development for autonomous vehicles is California’s Velodyne LiDAR, which uses time of flight (ToF) principles to measure the time it takes for IR light emitted from its LiDAR sensor to strike a surface or object and be reflected back to calculate the distance to the object.
As Wayne Seto, product line manager, explains: “The device can also 'sense' the intensity or reflectivity of the object surface, enabling it to provide a richer data set for 3D mapping and imaging, as well as autonomous driving and navigation. The sensor also rotates a full 360 degrees, and at a variety rotation angles, to generate a very accurate and detailed 3D point cloud.”
The company's sensors have already been used in a number of autonomous vehicles, including those under development by Ford. Although coy about the exact details, Seto also reveals that Velodyne is working with “other car makers and companies around the world to enable autonomous driving and ADAS in cars”.
Elsewhere, the Quebec company Phantom Intelligence has designed what it describes as a “compact and affordable” low-resolution solid-state LiDAR, initially aimed at ADAS applications. According to its president, Jean-Yves Deschênes, while a number of LiDAR solutions in the industry are aimed at fully autonomous applications – and the provision of high-density 3D maps of the environment at what he calls a “significant cost premium” – he argues that there is still a market for sensors that provide less complex information to process, while preserving the ability to make “critical safety decisions”. Many of the core opto-electronics components used in its platforms are manufactured by OSRAM, a long-standing supplier of such components for the automotive industry.
Although this system has not yet found its way into autonomous cars, Deschênes reveals that the company plans to integrate its systems in self-driving vehicle demonstrations by the end of 2017.
"We are believers in the progressive adoption of autonomy features. Our sensors will be in production for warning systems and data logging systems, not tied to any autonomous features, in early 2017. The time for the industry to validate those features into the heavily liability-laden ADAS functions will bring us to production no sooner than 2019. And the introduction into autonomous production vehicles will follow by the time the industry goes ahead with it," he adds.
Looking ahead, Deschênes points out that one of the key current hurdles to the adoption of LiDAR technology for autonomous vehicles is the manufacturability of the devices themselves. In his view, current scanning devices are “way too complex” to be efficiently and economically manufactured in automotive production quantities and he claims that the entire industry is now focusing on the “more repeatable and highly manufacturable potential of solid-state solutions”.
Other concerns for Deschênes relate to the robustness of the detections made by existing LiDAR devices, in particular how to ensure that no “false positive” readings are made, and the fact that LiDARs must “also be designed to operate under harsh weather conditions to be universally accepted”.
"The autonomous vehicle will be a multi-sensor vehicle. No technology out there does in all, reliably and under all circumstances. The industry is now just learning how to intelligently use sensors in fused-sensor environments, beyond just using the complementary sensors for redundancy purposes. The richness of information that can be obtained by using the complementary data from sensors is just being tapped. That understanding needs to be improved still," he says.
Meanwhile, Seto confirms that he is also aware of the ongoing concerns about the affordability of LiDAR systems for autonomous vehicles but argues that this challenge “will be overcome” once the technology is widely adopted and implemented.
"This will help us achieve the economies of scale [so] that it is affordable in every car. The market is already moving in that direction and Velodyne has plans in place to be ready to meet those needs," he adds.
17 May 2016 - 18 May 2016, Munich, Germany
As automated driving hits the headlines of the mainstream media, we take a step back and assess the challenges and opportunities that advanced driver assistance systems face now and in the imminent future.