With this background, we are currently conducting research on conditional cooperative control of robots. In particular, we focus on mutual prediction as a means of obtaining information necessary for control, and are developing a theoretical framework for the conditional cooperative control based on it. The video shows an experimental result obtained using a control method currently under development. When the paths of the two robots cross, one adjusts its moving direction so that it does not interfere with the other robot. The following is our publication related to this research.
日本語/English
Control Theory and Its Applications
Control theory is a framework for the modeling, analysis, and design of dynamical systems using mathematical tools. This has various applications in, e.g., engineering, physics, and biology. Through the studies on control theory and its applications, we address some real-world problems and work toward the creation of new values.
Research Interests
Modeling, Analysis, and Control of Networked Systems
In networked systems, subsystems are connected through networks.
These systems include modern applications such as swarm robots and sensor networks, and thus have been actively studied in the control community.
Networked systems are also found in the field of swarm intelligence, which highlights their importance as a research topic.
We model systems in various fields as networked systems and develop theories for their analysis and control that take into account the specific characteristics and objectives of each system.
Examples are given below.
- Formation control of multi-robot systems
- Distributed signal processing over sensor networks (in particular, distributed spatial filtering (DSF))
- Modeling and analysis of syncronized behavior of metronomes
- Analysis and control of power systems
Conditional Cooperative Control of Robots
Existing studies on multi-agent control typically assume that all agents work toward a common objective, such as achieving consensus. Meanwhile, high-performance robots are becoming increasingly accessible to the general public. These robots are expected to be used directly in industrial applications due to their advanced capabilities, which offer a promising solution to labor shortages caused by declining birth rates and aging populations. An example of their use in agriculture can be found in the reference. If high-performance robots become widely deployed in the future, multiple robots with different functions and roles may perform their own tasks independently within the same space, and cooperate only when necessary.
- S. Izumi and Y. Kawakita: Decentralized Priority Assignment for Two Robots based on Mutual Predictive Control, 2026 SICE Festival with Annual Conference (SICE FES 2026), Yokohama, September 14–17, pp. XXX–XXX, accepted
Achievements
Please see our journal and conference papers from the Publications page. Our achievements other than the papers are shown as follows.
Multi-Agent Mass Games
We developed distributed controllers for multi-agent mass games in which agents achieve formations displaying given images as shown in the following figure. Potential applications of mass games are entertainment, image processing, and display devices.
The following animation is a simulation result by our controllers, where the black dots indicate agents. What do you see?
The following images (shown in the right panel) are given to the agents: 1 2 3 4 Default
We developed a MATLAB simulator for mass games, as shown in the following paper. Our simulator can be downloaded as a MATLAB application installer from GitHub for free. A demonstration video of the simulator can be seen as the graphical abstract of the paper.
- S. Izumi, Y. Shiomoto, and X. Xin: Mass Game Simulator: An Entertainment Application of Multiagent Control, IEEE Access, Vol. 9, pp. 4129–4140 (2021) Open Access