A Second Autonomous Vehicle patent from Apple Surfaced Last Week Focused on Color Vision for Nighttime Driving
On Saturday Patently Apple posted a report titled "Apple reveals a Vital Safety System for Autonomous Vehicle's focused on eliminating Blinding Light Glare." Apple noted in their patent filing that "In an autonomous vehicle, the accuracy and reliability of a vehicle vision system is extremely important for safety because the autonomous vehicle may not have a mechanism for a human operator to override the autonomous vehicle maneuvers." The patent filing focused on light glare systems that could avoid problems for the vision system. Apple's second patent filing on vision systems focuses on color filter arrays for autonomous vehicles having such cameras.
According to the filing, machine visions systems are being used for an increasing number of applications, for example, in driver assistance systems for vehicles such as cars. These include backup cameras, lane change assistance, and parallel parking assistance systems. As vehicles progress toward fully autonomous operation, it is important to develop reliable machine vision systems that operate properly in many different conditions, for example, lighting conditions, weather conditions, traffic conditions, and use environments.
One of the most difficult operating conditions for a machine vision system is nighttime operation, for example, driving at night. This is because the low light levels make it difficult for a machine vision system to function properly at night.
Certain distinctions that can be made by the human eye at night, for example, on a dark street with glare from a street light or reflection off of a street sign, are more challenging for a machine vision system. In order to achieve widespread adoption of fully autonomous vehicles, however, it is important that the machine vision system is reliable for nighttime driving.
Apple's invention relates to a machine vision system includes a first camera configured to be coupled to a vehicle. In some embodiments, the camera can have an optical stack including a color filter array having a plurality of sections. In some embodiments, each section can include a first white filter portion, a yellow filter portion, and a magenta filter portion.
In some embodiments, each section can include a second white filter portion. In some embodiments, the first white filter portion and the second white filter portion can be disposed diagonally with respect to each other. In some embodiments, the machine vision system can be configured to provide information to an autonomous vehicle to operate the autonomous vehicle along a path, for example, a roadway.
In some embodiments, the optical stack can include a lens and an image sensor. In some embodiments, the image sensor can include a plurality of pixels. In some embodiments, each pixel can have a section of the color filter array disposed over the pixel. In some embodiments, the optical stack can include, in order, a lens, an infrared filter, a microlens layer, the color filter array, and an image sensor.
In some embodiments, the machine vision system can include a central processing unit coupled to the first camera. In some embodiments, the central processing unit can be configured to analyze an image captured by the first camera. In some embodiments, the central processing unit can be configured to identify at least one of a vehicle, a street light, a traffic sign, and a roadway marking in the image captured by the first camera. In some embodiments, the central processing unit can be configured to analyze the image captured by the first camera using a noise difference optimized color space.
In some embodiments, the central processing unit can be configured to analyze the image captured by the first camera using an automotive reference color chart. In some embodiments, the first camera can be configured to operate in lighting conditions of less than 1 lux.
Apple's patent FIG. 9 presented above illustrates a color space for human vision applications; FIG. 10 illustrates a color space for automotive machine vision applications; FIG. 16 illustrates an image processing chain.
Back in November Patently Apple posted a report titled "Apple Research Paper Touts VoxelNet as being Superior to LiDAR Regarding Autonomous Vehicle 3D Detection Methods," with images noted above.
The more you learn about Apple's deep R&D into autonomous vehicles the more you understand Tim Cook's statement that Project Titan is the 'Mother of all AI Projects.'
If you're an engineer involved with Machine Learning you're bound to appreciate Apple patent application, but for most, it's a difficult read. If you're curious enough, then review Apple's patent application 20180088588 here.
For interest sake, the two inventors of this particular patent don't seem to have a direct connection to Apple. The Inventor names support two outside engineers. The patent is noted in other USPTO documents to have been assigned to Apple via a NY Law Firm. The only Lucian Ion I was able to find was the Director, Camera R&D at IMAX in Toronto Canada. Whether the Lucian Ion we're highlighting is the Ion of this patent can't be confirmed outright at this time.
Prior to working at IMAX, Ion worked At Teledyne DALSA, a Canadian company specializing in the design and manufacture of specialized electronic imaging components (image sensors, cameras, frame grabbers, imaging software) as well as specialized semiconductor fabrication (MEMS, high voltage ASICs). Lucian Ion was the Director of Technology, Digital Cinema.
The second inventor listed is Mahesh Krishnamurthy from Saratoga California who has a history going back to India – with no direct link to Apple found. If anyone knows of a direct connection between these inventors and Apple that could be verified, please send in your information in our comment section.
Patently Apple presents a detailed summary of patent applications with associated graphics for journalistic news purposes as each such patent application is revealed by the U.S. Patent & Trade Office. Readers are cautioned that the full text of any patent application should be read in its entirety for full and accurate details. About Making Comments on our Site: Patently Apple reserves the right to post, dismiss or edit any comments. Those using abusive language or negative behavior will result in being blacklisted on Disqus.