Automotive ranging is facing a technical bottleneck, CMOS image sensing technology to help
when the current car can give an alarm to obstacles when reversing, give you directions, and automatically Park in parallel, you may think that the machine vision system is perfect. But this is not the case
in fact, only a small part of the vision system technology is used in all the above car functions. More new CMOS sensor chips are emerging. They can directly sense the spatial depth through 3D point-to-point, so that the machine vision system can detect the target and make the correct response
automotive 3D cameras are emerging
according to ABI research, the current automotive sensor market has reached a scale of 2.5 billion US dollars. According to frost Sullivan, the automatic Imaging Association and Piper Jaffray, the total sensor market in the fields of security, industrial automation and video games has also reached 750million US dollars
current automotive systems calculate the distance through ultrasonic sensors, for example, they can alarm obstacles when the driver backs up. At present, the more precise and expensive sensors under development use radar and lidar to achieve this function, but all companies in the world are trying a cheaper method - using cameras to calculate distance. The field of digital cameras has achieved scale, including Bio based composites based on polylactic acid materials; It can realize the conversion of agricultural straw into treasure, which makes it compete with analog sensors in costDavid Alexander, senior analyst of ABI research, pointed out: "people in the whole industry want to use digital cameras to replace more expensive ultrasonic, lidar and radar sensors. First, the cost of digital cameras is so low. Second, different driver assistant functions can use the same camera or even the same special computer chips and software to calculate distance."
machine vision algorithms and 3D cameras can be widely used in various metal, non-metal, composite materials, medicine, food, wood, copper, aluminum, plastic profiles, wire and cable, paper, film, rubber, textile, aerospace and other industries to test tensile performance indicators. Now there are many functions, including collision avoidance, lane departure alarm and lane keeping (so that you return to the original Lane), backward obstacle alarm, pedestrian monitoring Distance monitoring (keep your car at a proper distance from the car in front), night vision, adaptive headlighting, traffic/speed limit sign recognition, blind spot monitoring, etc
alexander pointed out: "we have a 2D camera for lane departure alarm: we can look in front of the car, follow the lane line, look around and identify the speed limit signs, monitor the blind spot area and give an alarm when a car is about to overtake your car. The new function can realize the crash accident before the crash, estimate the possibility of the crash, pre load the brakes and activate the airbag."
explore Canesta's CMOS image chip scheme
Canesta has been committed to developing CMOS image chips since its establishment in 1999. After seven years of efforts, the company has finally developed this most promising technology. Canesta's CMOS image chip can enable more and more enterprises to join the field of new energy vehicles, which can simultaneously calculate each point on the image chip through the time-of-flight mode in hardware, so as to sense the distance between the car and each object, rather than the ultrasonic sensor, which can only sense the distance between the car and the nearest object
alex's following data are all indicators of the zigzag experimental machine. Ander said: "Canesta's scheme must have a good prospect, because it uses only one camera and is based on CMOS technology, which is crucial to maintaining low cost. In addition, it uses the time-of-flight calculation method, which can bring a variety of functions to the car."
the time of flight formula calculation method refers to Canesta, which refers to the time when an infrared light source is used to illuminate the area with invisible light, and then the light is measured from the transmitter (behind the imager) to the outside object and finally returns to the CMOS detector. By using the hardware device on the CMOS chip, the machine vision algorithm can calculate the distance between each point in the area and the car, so as to easily classify various objects - in essence, this is accurate calculation, not just induction
jim spare, President and CEO of Canesta, said, "we use a standard 0.18 micron CMOS process that can be achieved by any factory. In addition, our sunshield technology allows us to sense the depth of objects without being affected by lighting conditions, rather than using an adaptive algorithm in the middle."
sunshield can sense every point and truly synchronously sense the difference between the ambient light before and after every 100 microseconds, unlike the adaptive software algorithm, which takes several milliseconds to adapt to the changing light environment
alexander also pointed out: "One of the most unique features of Canesta is its sunshield technology, which solves the biggest problem of using cameras - whether it can bring enough safety to the car in the changing light environment. I think Canesta is indeed aware of this, but they still have a lot to do to solve various problems in car functions. I think Canesta may launch its first model in 2008 or 2009 Car 3D camera. "
Canesta's first product in the automotive field may be the 3D camera used in the new Honda, which is installed in the driver's seat to sense the shape of passengers and control airbags. Now this function is realized through the pressure pad on the seat, but the cost of using a CMOS camera may be as low as the pressure pad, but it can bring many other functions to the car, including giving an "anti sleep" alarm - distinguishing the difference between the normal blinking of the driver and the beginning of eye closure due to insufficient sleep
other companies are also developing CMOS image chips that can detect the depth of objects point-to-point, including International Electronics Engineering S.A. Mobileye, another company, did not use a 3D imager, but chose an ordinary 2D camera combined with its CMOS hardware accelerator chip to analyze and calculate the distance