Four-legged Robot Autonomously Explores Hazardous Tunnel Construction Areas, Realizing Remote Observation
Konoike Construction and Pocket Queries successfully conducted a demonstration experiment in which a four-legged robot autonomously performed observations and acquired environmental data in hazardous mountain tunnel construction zones.
📋 Article Processing Timeline
- 📰 Published: May 26, 2026 at 00:20
- 🔍 Collected: May 25, 2026 at 16:01
- 🤖 AI Analyzed: May 25, 2026 at 16:03 (1 min after Collected)
## Four-legged robot enters hazardous areas
Konoike Construction Co., Ltd. (Headquarters: Chuo-ku, Osaka; President: Hiromi Watatsu) and Pocket Queries, Inc. (Headquarters: Shinjuku-ku, Tokyo; CEO: Nobuhiko Sasaki) jointly conducted a demonstration experiment using autonomous four-legged robots on February 8, 2026, aimed at improving safety during observations of hazardous areas in mountain tunnel construction. This experiment confirmed the possibility of unmanned exploration in dangerous zones and the acquisition of environmental information such as gas detection and point cloud data.
## Hazards in tunnel construction
In mountain tunnel construction, observing the excavation face is essential to understand and record the state of the ground. However, this work involves various risks such as rockfalls, collapses, and the emission of flammable gases. To address this issue, Konoike Construction and Pocket Queries have been developing an observation system using four-legged robots to ensure worker safety while performing accurate observations.
## How the robot moves: The process of autonomous walking
First, the robot is activated at a launch base installed approximately 60 meters from the deepest point of the tunnel excavation, and observations are conducted in the following steps:
- Move from the launch base to the start point, scan the surrounding environment, and record position information of itself and obstacles as three-dimensional spatial data.
- Analyze the spatial data and begin autonomous walking toward the tunnel tip while avoiding obstacles.
- The robot automatically stops at a preset point to observe the excavation face and collect environmental data.
- After observation and data collection, the robot returns to the launch base by following the same path, referring to the spatial data recorded on the outbound trip.
The launch base is equipped with a charging function, allowing continuous operation by charging the robot between observation tasks.
## Onboard equipment
For the demonstration experiment, the system was based on a Unitree B2-W four-legged robot, enabled for autonomous walking by detecting the surrounding environment, and equipped with functions such as remote excavation face observation via camera and environmental monitoring via sensors.
- Sensor system: 3D LiDAR (for spatial recognition, 3D environmental mapping, and obstacle detection), gas sensors and dedicated terminals (for detecting/measuring CH4, O2, H2S, CO, and CO2 concentrations).
- Imaging/Video system: Gimbal camera (for high-precision photography), POV camera (for remote operation video recording).
- Control/Processing system: Control PC (for integrated processing of various sensor data and movement control).
- Communication system: Communication module (for remote control/monitoring), wireless LAN connectivity.
- Safety/Display equipment: Flash indicator lights (to improve visibility to surroundings).
## Experimental results: Confirmation of balance between precision and safety
In the experiment, the robot was set to stop 15 meters before the deepest point of the tunnel construction, and observations of the excavation face were performed after stopping at the target point. Through remote observation, weathering, water seepage, and gas detection were confirmed. After data collection, the robot returned along the same path as the outbound journey. The results showed that it is possible to improve worker safety and perform comprehensive environmental evaluation while maintaining accuracy comparable to conventional manual visual work.
## Future outlook
Going forward, to address issues such as surveys during ground collapse, the companies will work on validating applicability in more complex terrain conditions and ensuring stability during long-term operations, aiming for the commercialization of this system. Through this initiative, Konoike Construction and Pocket Queries contribute to the improvement of safety and efficiency in mountain tunnel construction.
Konoike Construction Co., Ltd. (Headquarters: Chuo-ku, Osaka; President: Hiromi Watatsu) and Pocket Queries, Inc. (Headquarters: Shinjuku-ku, Tokyo; CEO: Nobuhiko Sasaki) jointly conducted a demonstration experiment using autonomous four-legged robots on February 8, 2026, aimed at improving safety during observations of hazardous areas in mountain tunnel construction. This experiment confirmed the possibility of unmanned exploration in dangerous zones and the acquisition of environmental information such as gas detection and point cloud data.
## Hazards in tunnel construction
In mountain tunnel construction, observing the excavation face is essential to understand and record the state of the ground. However, this work involves various risks such as rockfalls, collapses, and the emission of flammable gases. To address this issue, Konoike Construction and Pocket Queries have been developing an observation system using four-legged robots to ensure worker safety while performing accurate observations.
## How the robot moves: The process of autonomous walking
First, the robot is activated at a launch base installed approximately 60 meters from the deepest point of the tunnel excavation, and observations are conducted in the following steps:
- Move from the launch base to the start point, scan the surrounding environment, and record position information of itself and obstacles as three-dimensional spatial data.
- Analyze the spatial data and begin autonomous walking toward the tunnel tip while avoiding obstacles.
- The robot automatically stops at a preset point to observe the excavation face and collect environmental data.
- After observation and data collection, the robot returns to the launch base by following the same path, referring to the spatial data recorded on the outbound trip.
The launch base is equipped with a charging function, allowing continuous operation by charging the robot between observation tasks.
## Onboard equipment
For the demonstration experiment, the system was based on a Unitree B2-W four-legged robot, enabled for autonomous walking by detecting the surrounding environment, and equipped with functions such as remote excavation face observation via camera and environmental monitoring via sensors.
- Sensor system: 3D LiDAR (for spatial recognition, 3D environmental mapping, and obstacle detection), gas sensors and dedicated terminals (for detecting/measuring CH4, O2, H2S, CO, and CO2 concentrations).
- Imaging/Video system: Gimbal camera (for high-precision photography), POV camera (for remote operation video recording).
- Control/Processing system: Control PC (for integrated processing of various sensor data and movement control).
- Communication system: Communication module (for remote control/monitoring), wireless LAN connectivity.
- Safety/Display equipment: Flash indicator lights (to improve visibility to surroundings).
## Experimental results: Confirmation of balance between precision and safety
In the experiment, the robot was set to stop 15 meters before the deepest point of the tunnel construction, and observations of the excavation face were performed after stopping at the target point. Through remote observation, weathering, water seepage, and gas detection were confirmed. After data collection, the robot returned along the same path as the outbound journey. The results showed that it is possible to improve worker safety and perform comprehensive environmental evaluation while maintaining accuracy comparable to conventional manual visual work.
## Future outlook
Going forward, to address issues such as surveys during ground collapse, the companies will work on validating applicability in more complex terrain conditions and ensuring stability during long-term operations, aiming for the commercialization of this system. Through this initiative, Konoike Construction and Pocket Queries contribute to the improvement of safety and efficiency in mountain tunnel construction.
FAQ
四足歩行ロボットによるトンネル工事の実験の目的は何ですか？
山岳トンネル工事の危険エリア（落石、崩落、可燃性ガスの湧出など）において、作業員の安全を確保しつつ、正確に掘削面や環境情報の観察・取得を行うことです。
実証実験で使用されたロボットは何ですか？
Unitree社製の四足歩行ロボット「B2-W」をベースに、3DLiDAR、ガスセンサー、ジンバルカメラなどを搭載したシステムです。
ロボットはどのように自律歩行するのですか？
あらかじめ記録した三次元空間データを解析し、障害物を回避しながら目標地点へ移動します。帰還時も記録した経路をたどって発進基地へ戻ります。
実験でどのようなデータが取得されましたか？
ジンバルカメラによる掘削面の高精度な観察映像、およびCH4、O2、H2S、CO、CO2などのガス濃度データなどが取得されました。
今後の課題は何ですか？
より複雑な地形条件での適用性検証や、長時間運用時の安定性確保を行い、実用化を目指すとしています。