In our rapidly developing information-based society, fundamental research for not only physical object analysis but also massive data analysis and utilization, as typified by AI, has become extremely important.
In order to reveal the unknown rules behind data generated from various events such as natural phenomena, industrial activities, living behavior, and to produce new valuable information, we carry out advance research for data analysis.
Our research has potential applications in the field of intelligent systems, including automated driving system.

Decentralized Autonomous Coordination Technology

We are researching decentralized control theories to achieve intelligent coordination within the networks of infrastructure, mobility, robots, and sensors of future smart societies.

Demonstration of Decentralized Autonomous

Coordination Technology
This is a simulation by around 10,000 robots to reproduce a 12-color painting. Each robot carries a panel painted with a single color. The painting is completed when the robots line up in the locations determined for their color panel. Compared to the conventional decentralized control methodology shown in the center below, the proposed method at the bottom right is capable of simulating the painting through mutual coordination between the robot positions.

Demonstration of Decentralized Autonomous

Virtual Evaluation Techniques

With the aim of further boosting the reliability of automated driving and other technologies, we are researching virtual evaluation techniques capable of computerized systems testing.
More specifically, we are developing a technique that can generate virtual images that can be viewed from any angle (i.e., free-viewpoint images) through the application of computer vision (CV) technology based on data measured by vehicles capable of highly efficiently obtaining omnidirectional images and three-dimensional point groups. We can then simulate a wide range of phenomena, such as changes in the weather and different vehicle types through the application of computer graphics (CG) technology. This active fusion of CV and CG technologies enables us to carry out high-quality systems evaluations in a virtual space.

Fig1. Data measurement by experiment vehicle Fig2. Generation of virtual image by fusion of CV and CG