Since the Uber crash, developers of autonomous driving technology have tried their best to regain confidence, and the laser pointer thermal sensor has quietly become the focus of attention.
What kind of technology can avoid such a tragedy? Is the laser pointer and ultrasound? Still radar and camera?
The above problems have always dominated the major controversy in the field of autonomous driving technology. In the past, people have never been so concerned about overheating.
But with the serious accident in Uber, Arizona, people began to pay attention to thermal sensors.
At present, thermal imaging camera manufacturing giants such as FLIR Systems and AdaSky have become the focus of the public, and to promote their case to the public who are wary of the development of autonomous vehicles.
The focus of the autopilot technology debate: any sensor technology has inherent inherent advantages and disadvantages
According to reports, the self-driving cars of companies such as Waymo and Uber rely mainly on rotary lidar (LiDAR) sensors to accurately detect the surroundings of the vehicle. However, even though the price of top LiDAR has fallen dramatically over the past few years, it is still very expensive for most market applications. For example, Velodyne's 64-wire LiDAR system costs about $80,000.
Market-savvy developers have replaced them with cheaper laser pointer technology. The new generation of so-called solid-state LiDAR is much more compact than the rotating laser system on the Waymo. In the next few years, the price of solid-state LiDAR is likely to drop to less than $100.
The gradual reduction in LiDAR costs is not surprising, but it requires significant trade-offs in terms of cost and performance. Compared to high-resolution LiDAR, low-resolution LiDAR sensors are difficult to accurately distinguish distant objects, which means that the response time of these self-driving cars that rely on low-resolution LiDAR systems will be shorter.
Even more tragic is that LiDAR is a spectroscopy technique, and fog, rain and snow can seriously affect its accuracy.
Therefore, Tesla chose a very different sensing strategy: get rid of LiDAR and switch to radar and optical cameras. However, these two techniques also have limitations: although radar can sense distant objects and perform well even in inclement weather, it is difficult to identify objects.
In Tesla's system, objects are judged by working with cameras and radars. However, the camera does not produce well in low light, which is the Achilles heel of the system.
It is worth noting that according to the latest research from the Insurance Institute for Highway Safety (IIHS), the number of pedestrian deaths rose the fastest after sunset. In fact, the vast majority (three quarters) of pedestrian deaths occur at night.
After Uber's fatal accident, the police released a shocking video. Obviously, it can be seen from the video that the lighting conditions played a leading role.
In LiDAR-based or radar-based systems, thermal sensing technology may be a necessary expedient. The technology has been used in high-end cars such as BMW and Porsche, but it has not yet become a must-have for most autonomous driving sensors.
A recent white paper published by AdaSky said: "The new sensors using far-infrared (FIR) technology can fill the reliability gap left by other autonomous vehicle laser pointers. Because FIR is already used in defense, security, fire protection and construction. It has been a mature and proven technology for decades, and the company is one of the companies that use commercial infrared sensing for autonomous driving.
Using infrared waves, FIR-based cameras detect the different heat that the object naturally releases. With its ability to sense infrared spectroscopy, FIR cameras can detect objects that are not recognized by visible light cameras, radars, and LiDAR systems. Crucially, the FIR camera works well in low light and inclement weather conditions.
Given the public's response to recent auto-driving car accidents, developers will work hard to "how to regain public confidence and consumer enthusiasm." We will see more discussion on improving sensor packaging, and it is certain that thermal imaging cameras will become an important part of autonomous driving sensors.