"Camera-LiDAR Fusion Sensor"
[Click to enlarge image]
Dropping objects being recognized by the newly-developed sensor. Tires can be detected.
[Click to enlarge image]
Dropping objects being recognized by a conventional LiDAR sensor. Tires cannot be detected.
[Click to enlarge image]

Kyocera Corp exhibited "Camera-LiDAR Fusion Sensor," which integrates a camera and a LiDAR (light detection and ranging) sensor at Automotive Engineering Exposition 2018, which took place from May 23 to 25, 2018, in Yokohama City, Kanagawa Prefecture, Japan.

Kyocera aims to commercialize the new sensor in 2022 or 2023. The sensor uses two kinds of data collected by the camera and LiDAR sensor, respectively, to improve the recognition rate of objects around the sensor. For example, it recognizes three-dimensional (3D) objects such as humans and buildings by using images (taken by the camera) and intensively scans the environment surrounding the objects by using LiDAR.

"When only LiDAR is used, it is difficult to detect the locations of objects," Kyocera said. "When only the camera is used, there is a possibility that shadows on the ground are mistakenly recognized as three-dimensional objects."

When the camera and LiDAR sensor are combined, it becomes possible to recognize 3D objects about 15cm above the ground, enabling to detect falling objects such as tires, which is normally difficult, the company said.

Moreover, Kyocera eliminated the mechanical driving part of the LiDAR sensor by using a MEMS technology for beam scanning, instead of a motor, to improve durability and reduce size. The resolution of the LiDAR sensor is 0.05°. The company plans to apply the Camera-LiDAR Fusion Sensor not only to automobiles but also to robots and drones.

Original Japanese article