Research Focus
Cooperative Multi-Robot System for Automated Roadside Weeding
This study aims to develop a cooperative roadside weeding system using multiple small robots. Roadside weeding still relies heavily on manual labor, leading to labor shortages and heavy physical burdens on workers. However, concentrating sensing, decision-making, and weeding functions in a single advanced robot has limitations in terms of cost, scalability, and practical operation. To address these issues, this study proposes a system comprising three robots a leading robot, a working robot, and an inspection robot. These robots move in a line along the road shoulder, each playing a different role. The leading robot observes terrain conditions and detects obstacles ahead. The working robot uses this information to detect and remove weeds. The inspection robot follows behind, evaluates the work results, and requests additional weeding where necessary. By sharing position data and task status, the robots adjust their spacing, speed, and stopping and restarting timing. This approach is expected to reduce missed weeds and unnecessary movement, resulting in safer, more efficient outdoor weeding.
Autonomous Robot Navigation System for Semantic Outdoor Map Generation
This study aims to develop a cooperative autonomous navigation system for multiple R4 (Ritsumeikan Road Removal Robot) units by introducing a leading robot that acts as a command unit. The goal is to improve the operational efficiency of autonomous roadside weeding. The leading robot integrates color and depth information obtained from an onboard RGB-D camera to generate a semantic map in real time. This map enables both object identification and three-dimensional position estimation. In addition to obstacle locations, the map includes weed distribution density and terrain attributes. Based on this information, the system dynamically calculates and assigns optimal work areas and travel paths to multiple following weeding robots. The proposed algorithm enables the robots to adjust their tasks and routes in response to changes in the surrounding environment. By closely integrating advanced visual environment recognition with multi-agent cooperative control, this research aims to automate efficient weeding over wide roadside areas, which is difficult to achieve with a single robot.