This research and development initiative focuses on robotic technologies that can flexibly respond to various scenarios, with the goal of enabling effective recovery operations for river channel blockages. Below, we introduce the main achievements from fiscal year 2024.
For the installation of drainage pumps—a key part of the system integration for emergency river channel blockage—Kumagai Gumi has proposed a system using compact construction equipment, as shown in Figure 8. However, this approach may not be feasible in areas with unstable ground conditions. In response, a research group at Osaka Institute of Technology has developed a modular transport container system called “BRAINS.” This system allows the working unit and the mobile unit to be separated and reconnected as needed, making it easier to adapt to changing conditions at disaster sites. The container is divided into two levels: the upper level houses the working unit, while the lower level contains the mobile unit. By swapping out mobile units according to the site environment and recombining them with the working unit, a wide range of construction robot configurations can be realized. In 2024, the team successfully created a scale model of this concept and conducted functional tests in an indoor environment. Figure 9 shows an overview of the BRAINS container enabling the separation and recombination of the mobile and working units (top left), the d-FlexCraw mobile unit designed for rough terrain (top right), and the i-CentiPot Ammonite working unit used for hose deployment (bottom). In this figure, the hose-deploying working unit is mounted on a standard crawler-type mobile platform.

Meanwhile, for excavation work in earthmoving during the construction of temporary drainage channels, hydraulic excavators are currently used. To replace the excavators, we are conducting R&D on new excavation methods that would allow this task to be performed by multiple small construction robots. Specifically, we aim to establish an excavation technique using two compact robots, by dividing the excavation process into two functions: loosening and scooping. In 2024, we successfully automated each step of this process: one compact construction robot used a chiller (a ground-loosening tool, shown in the top left of Figure 10) to loosen the soil (top right), while a separate skid loader collected the loosened soil.

In addition to this, we also advanced research on dynamic cooperative control technologies for multiple construction robots working together to respond to river channel blockages. Furthermore, we developed AI-based learning methods for excavation strategies that can adapt to varying ground strengths.