In recent years, robots have become more and more sophisticated. The main reason for this is that the processing speed of ICs is rapidly increasing, and thanks to the expansion of sensor types and their uses, the variety of tasks that robots can do are increasing explosively.
There are many different types of robots, but if we define a robot as a “replacement for a human being,” the equivalent of an eye is a camera. In this section, we will explain how artificial solar lighting contributes to the development of robot-mounted cameras.
The accuracy of image recognition has been greatly improved by the increased performance and processing speed of cameras. For example, facial recognition uses AI to detect whether a person’s face is recognized or not, and not only facial features are used for detection, but also the irises of the eyes and the skeletal frame.
To perform facial recognition, the computer on board the robot must accurately recognize people’s faces even in backlit scenes or in low light. This task is handled by image processing software, and not by the camera.
When the sun is behind a person, a backlit condition with the largest luminance difference occurs. If you want to conduct development tests under this backlit condition, you regularly get this kind of environment outdoors during the day, but it is also affected by the weather. In addition, the light environment changes from moment to moment.
Therefore, it is extremely important for manufacturers to be able to reproduce and maintain poor light conditions to ensure a good test environment. In addition, it is also very important that the light quality has spectral characteristics similar to that of sunlight, rather than just being bright.
It is very important for manufacturers to put in place such a light environment in order to improve the accuracy of image recognition, and artificial solar lighting is playing an active role in this situation.
The photos above show how people’s faces change in different light environments. If there is a large luminance difference in the angle of view, as shown in the photos above, it will be difficult to capture the object accurately.
The camera calculates the luminance difference for each pixel from the captured image, and recognizes the object by amplifying the pixels with low luminance, but the amplified data also results in too much noise. Therefore, soft processing such as noise canceller is required.
A stable light environment is essential for an accurate evaluation of whether this software is properly processing the data. One of the characteristics of artificial solar lighting is that the light does not change, and the change it undergoes over time is extremely small. This characteristic is one of the reasons why we are highly regarded by our customers.
The model mainly chosen by our customers is the XELIOS XG-500AF artificial solar lighting. Customers who require both high luminance and high illuminance use the super spotlight lighting type XELIOS XG-500AFSS.
Many robot manufacturers, camera manufacturers, and image processing software manufacturers use XELIOS artificial solar lighting, and it greatly contributes to the construction of image processing systems for robots.
Refer to the pages below for specifications and performance.
Artificial solar lighting demo machines are also available. You can also try them out in advance at the actual test site. If you are interested in a light source for developing robot-mounted cameras, please feel free to contact our experienced team.
