Thermal imaging cameras can detect small temperature differences as low as 0.1 degrees Fahrenheit at 100 meters
Not long ago, Uber had a fatal autopilot vehicle accident in Tempe, Arizona, USA. Most observers had two basic questions about the accident: Why did the self-driving vehicle not see the dead passing through the street? And stop before the collision? How can we prevent this from happening again?
The shared car service company has suspended the autonomous vehicle test program indefinitely and is working with the NaTIonal TransportaTIon Safety Board (NTSB) to conduct an accident investigation. The NTSB has not disclosed any findings, but LiDAR (a laser radar, a laser sensor that detects and identifies the deceased even in complete darkness) on an autonomous vehicle has become the focus of the accident. It is possible that the LiDAR has a blind spot to detect, or lacks the resolution to identify the deceased as a pedestrian, or the car's software algorithms are not able to translate these detected data point clouds into the correct vehicle decisions, thereby taking emergency braking or bypassing the deceased.
Infrared night vision and target recognition function for BMW 7 Series sedan
In any case, the accident raised a life-and-death debate that made people pay attention to any method that would make the autonomous driving system safer. This has also brought new market opportunities for infrared manufacturers – equipped with auto-driving vehicles with thermal imaging cameras that meet the requirements of the car.
"Infrared cameras are unique in detecting some of the targets you least want to collide," said Mike Walters, head of FLIR's product division. "There is no doubt, of course, people." Traditional cameras require visible light to "see" the world. FLIR's thermal imaging sensors primarily utilize far-infrared spectroscopy. They are capable of detecting small temperature differences as low as 0.1 degrees Fahrenheit, so even in the cold winter night, the slim bicycle wheel can be clearly displayed, and its detection range is more than 240 meters. This detection distance is the most performance on the market. A few strong LiDARs are comparable. In addition, unlike LiDAR systems, infrared thermal imaging sensors are free from fog or direct sunlight.
According to a report by Mammus Consulting, FLIR released a new high-resolution thermal imaging vehicle development kit (ADK) at the beginning of this year. The ADK is equipped with a high-resolution FLIR Boson movement and is equipped with Intel Movidius Myriad 2 vision. The processing unit is FLIR's only infrared camera module that matches the low-power multi-core vision processor in a compact, rugged package that meets the vehicle's specifications.
Such thermal sensors have been widely used in many fields: heat tracking missiles, circuit problem detection, and location determination of personnel at fire sites. Of course, FLIR is not the only vendor that strives to make infrared thermal imaging sensors compliant with vehicle standards. “We enable infrared thermal imaging sensors to have the ruggedness and durability required for automotive applications; reduce their power consumption to enable them to be used in smartphones; improve their seismic calibration performance for gun sights; Thermal performance allows them to cope with high temperatures such as combustion," said TIm LeBeau, head of corporate strategy at Seek Thermal, a FLIR competitor in Santa Barbara. Therefore, it is not difficult to install an infrared camera into a vehicle in the face of a bumpy outdoor driving environment.
According to a report previously reported by Memes, Seek Thermal exhibited its first high-resolution infrared thermal imaging camera for the automotive aftermarket at CES 2018 at the beginning of the year, featuring a 320 x 240 pixel high resolution thermal imaging sensor. The 24-degree field of view is equipped with a two-element sulfur-based lens that is easily integrated with existing in-car entertainment systems and is priced at less than $999.
FLIR has applied machine learning technology to infrared readout to help computers identify pedestrians and cyclists
In fact, FLIR has supplied infrared thermal imaging sensors, such as the BMW 7 Series luxury sedan, for hundreds of thousands of vehicles over the past decade. In the dark nights when animals or people are difficult to observe, the driver can see any shape obstacles such as bright white human bodies or deer on the large car screen through the infrared camera.
The first step is to realize the assembly of the thermal imaging camera on the vehicle. The next logical improvement is that FLIR is practicing it. It is to train the on-board computer to identify the obstacles detected by the thermal imaging camera. FLIR has applied machine learning technology to infrared readouts to help computers learn to identify objects such as pedestrians and cyclists, just like other traditional visible light camera manufacturers are researching and applying image recognition technology. FLIR hopes to create a system that can use thermal imaging to automatically determine the condition in front of the vehicle to alert the driver or take emergency braking if necessary.
For such driving assistance functions, it is easy to design an infrared thermal imaging sensor for a fully automated vehicle. However, it is not intended to replace the current radar, visible light camera or LiDAR, but another environmental sensing and target detection solution, which is a powerful complement to other technologies.
“One of the core strengths of infrared imaging is its natural complementary nature,†said Karl Iagnemma, current CEO of Nutonomy, a US automaker, acquired by auto supplier Delphi. “They can be used in dark environments and bad weather conditions. Stable work."
Of course, such thermal sensors are not the perfect technical solution. First of all, they are not cheap. For example, Seek Thermal's thermal imaging camera costs about $2,500, but the company is currently working to reduce costs to around $1,000. Compared to traditional visible light cameras, this is simply expensive (which is why we only see infrared night vision on the BMW 7 Series million-class luxury car). Of course, compared to LiDAR sensors, it is still cheap.
In addition, Iagnemma added that the resolution of the thermal imaging camera is not as high as that of a conventional visible light camera. Moreover, infrared light waves cannot penetrate transparent glass, so there is no potential threat that the camera can "see" behind the glass. However, as mentioned earlier, thermal imaging cameras are a powerful addition, and those conditions are not the same as traditional cameras.
Marine Control Cable,Cwb Marine Control Cable,Ship Control Cable,Shield Marine Cable
Baosheng Science&Technology Innovation Co.,Ltd , https://www.bscables.com