事例紹介cases

Toyoda Institute of Technology Smart Vehicle Research Center​ ​​ ​

Vehicle with Robovision2 installed
Toyoda Institute of Technology Smart Vehicle Research Center is promoting R & D on driving assistance systems that are safe for elderly people and beginner drivers and have less burden on the environment. Specifically, the computer detects various objects such as a pedestrian, a vehicle, a road, a white line, and the like based on information obtained from a sensor such as a TV camera, a laser scanner, a GPS (global positioning system).We are developing technologies that can recognize the running environment. In this research and development, RoboVision 2 is utilized as a tool for developing nighttime Image Recognition technology with cameras.
As a study of "stereo technology robust to changes in driving environment" in recent years, we are studying how to construct a recognition system that is resistant to changes in the driving environment such as snowfall, rain, nighttime, fog.We are introducing the results of investigation below.

Study on stereo technology robust to changes in driving environment

1. Research background
As a demand for stereo vision in a vehicle,
 ・周辺環境の光の変化
 ・振動などによるカメラのアライメントのずれ
 ・リアルタイム処理
 ・細い支柱やチェーンなどの小さい物体の検出
 ・奥行情報の取得
There are various demands, and in response to that request, Professor Mita's research is carrying out various initiatives and introduces some of the efforts below.
2. Auto calibration function
We are developing algorithms to automatically calibrate by calculating parameters to correct image distortion from multiple images in the research against changes in hardware factors of the camera being used. Below is the distance measurement result comparing before and after auto calibration.
By reflecting the auto calibration parameters, it is possible to obtain distance measurement results in a wider area.
3. Efforts to improve visibility at night
Image noise increases in nighttime images. Although we will skip the calculation algorithm, we have succeeded in acquiring parallax images with less noise by calculating the parallax after removing the noise of the image by referring to the surrounding information of the matching point. Moreover, by using RoboVision 2 equipped with a high sensitivity CMOS sensor, high brightness image measurement is possible even at night.
Distance measurement result at night
4. Stereo vision system in tunnel
For use as an in-vehicle camera, images may be blurred in tunnels and it may be difficult to calculate accurate depth information. Especially, when the speed is high or the distance to the measurement target is close, the tendency tends to be noticeable.
Measurement image examples in the tunnel
In this research, in order to solve the image blur in the tunnel, we are developing a method to reduce the blur caused in the tunnel by considering the speed of the vehicle and the shutter speed of the camera.
Comparison of distance measurement results in tunnel
5. Calculation of depth information using deep learning
As other efforts, there is explanation on Image Recognition technology making use of deep learning, and semantic segmentation, lane detection, and Image Recognition results combining these are introduced.

​ ​Image recognition and Image processing products

RoboVision 2s
Super high sensitivity stereo camera system compatible with ranging and object detection software
RoboVision 3
A stereo camera capable of sensing a distance of up to 150 m, horizontal 100 ° and field of view
To TOP