Smart Robot Patrols Coal Mine Airway in China
In a move that signals a new phase in the automation of underground mining operations, engineers and researchers from Shaanxi Coal Group’s Shenmu Zhangjiamao Mining Co., Ltd. have deployed an intelligent inspection robot system designed to autonomously monitor one of the most hazardous zones in a coal mine: the return airway. The innovation, detailed in a recent publication in Safety in Coal Mines, showcases a fully integrated robotic solution capable of navigating long, complex underground tunnels, collecting real-time environmental data, identifying structural anomalies, and even detecting early signs of flooding—all without requiring human presence in the high-risk zone.
The development of this system addresses a longstanding challenge in the mining industry: the need for frequent, reliable inspection of return airways, which are critical pathways for ventilating toxic gases and smoke in the event of a fire or explosion. These tunnels, often stretching over five kilometers in length, are not only physically demanding to traverse but also pose significant health and safety risks due to poor visibility, potential gas accumulation, and structural instability. Traditionally, these inspections have been carried out manually, relying on workers to walk the length of the tunnels, using visual, auditory, and olfactory cues—commonly referred to as “look, listen, smell, measure”—to detect anomalies. This method, while still in use at many mines, is inherently inefficient, inconsistent, and exposes personnel to unnecessary danger.
The new robotic system, led by Mao Hao, a senior engineer and Deputy Director of the Intelligent Office at Zhangjiamao Mining, represents a significant leap forward in mine safety automation. According to the research team, the robot is not merely a remote-controlled camera on wheels; it is a fully autonomous, self-powered, and self-navigating unit designed specifically for the harsh, unstructured environment of an underground coal mine. The project, which was implemented in the 2-2 coal seam return airway at Zhangjiamao, integrates advanced robotics, sensor fusion, wireless communication, and machine learning into a single, cohesive platform.
At the heart of the system is a cable-dragged robotic unit suspended from an overhead steel cable that runs the entire length of the monitored tunnel. Unlike free-roving robots that rely on onboard navigation systems such as LiDAR or GPS—which are ineffective underground due to signal blockage—the Zhangjiamao robot uses a tethered, rail-free design. The robot is pulled along the cable by a drive mechanism located in a safe, non-hazardous area—the electromechanical pump house—eliminating the need for electrical equipment in the explosive atmosphere of the return airway. This design adheres strictly to coal mine safety regulations, which prohibit the use of explosion-prone electrical components in such zones.
The drive system employs a continuous loop of steel wire rope, guided by a series of pulleys and tensioning mechanisms. The robot attaches to the moving cable via a clamping device known as a “gripper,” allowing it to travel back and forth along the designated route. This setup enables the robot to cover long distances reliably and repeatedly, with minimal mechanical wear and high positional accuracy. The movement is controlled by a central command system that monitors the robot’s location in real time using a high-precision rotary encoder. This encoder measures the rotation of the drive wheel, allowing the system to calculate the robot’s exact position based on speed and time—a method known as time-velocity mapping. To correct for cumulative errors caused by slippage or sensor inaccuracies, the system incorporates periodic position calibration points along the route, ensuring centimeter-level accuracy over kilometers of travel.
One of the most innovative aspects of the system is its ability to pass through ventilation doors—critical safety structures that control airflow and prevent the spread of contaminants during emergencies. Because the robot and its cable must cross these sealed doors, the engineering team designed a specialized dual-window mechanism. Two small, motorized wind flaps are installed on either side of each door. As the robot approaches, photoelectric sensors trigger a sequence in which the downstream flap closes first, maintaining the pressure differential, before the upstream flap opens to allow the robot to pass. Once through, the process reverses, ensuring that the integrity of the ventilation system is never compromised. This automated door-crossing protocol is a key enabler for long-range inspection in segmented mine networks.
Powering the robot presented another significant challenge. Since the return airway cannot host charging stations or power lines, the team opted for an intrinsically safe (IS) battery system housed within the robot itself. This battery, designed to prevent sparks or overheating even in the presence of flammable gases, provides enough energy to operate the robot’s sensors, cameras, and communication systems for 8 to 9 hours on a single charge. When the battery level drops below a user-defined threshold, the robot automatically aborts its current mission and returns to the pump house, where it docks with a wireless charging station. The system includes fail-safes: if charging fails, the robot triggers an alert for manual intervention. Additionally, a manual charging port is available for emergency use, ensuring operational continuity even under adverse conditions.
The robot’s sensory suite is comprehensive. Mounted on the front and sides are high-definition cameras capable of panning and tilting to capture detailed visual data from multiple angles. These cameras feed into an onboard data processing unit that runs advanced image analysis algorithms. One of the standout features is the system’s ability to detect water accumulation—a common but dangerous issue in aging mine tunnels. Using a combination of digital image processing and a deep learning model based on Faster R-CNN (a region-based convolutional neural network), the system can identify puddles, calculate their surface area, and assess their potential risk. The model was trained on an augmented dataset that included grayscale conversion and angular transformations to improve recognition accuracy under varying lighting and camera angles. When water is detected, the system automatically alerts the control center, logs the event, and stores the image for further analysis.
Environmental monitoring is another core function. The robot is equipped with a multi-parameter sensor array that continuously measures gas concentrations—including methane, carbon monoxide, and oxygen levels—along with air velocity, temperature, humidity, and dust density. These readings are transmitted in real time via a dedicated Wi-Fi network installed along the tunnel at 400-meter intervals. Each access point is an intrinsically safe wireless base station, ensuring reliable communication even in the presence of electromagnetic interference or signal attenuation from rock formations.
To ensure data integrity, the system employs a dual-communication strategy. While live data is streamed to surface control rooms for immediate monitoring, all sensor readings and video footage are also stored locally on the robot’s internal memory. If the wireless link is interrupted—due to equipment failure or signal obstruction—the data is preserved and uploaded later when the robot returns to the pump house. There, it connects to the mine’s existing 4G network to synchronize with central databases. This redundancy guarantees that no critical information is lost, a crucial feature for compliance and incident investigation.
From a human-machine interaction standpoint, the system is designed for ease of use. Operators on the surface can view the robot’s live feed, issue commands, and review historical data through an intuitive graphical interface. The system also supports remote control from underground terminals, allowing supervisors to conduct virtual inspections without entering the hazardous zone. Alerts and warnings are displayed prominently, and the software includes tools for generating trend reports, such as gas concentration over time, which aid in predictive maintenance and risk assessment.
The practical impact of the system has been substantial. Since its deployment, the need for manual inspections in the 2-kilometer test section has been reduced by over 80%. The robot performs multiple rounds per day, collecting more data than a human inspector could in a week. More importantly, it does so without exposing personnel to the risks of gas exposure, roof falls, or entrapment. The system has already detected several minor issues—including a small methane leak and early signs of wall deformation—that might have gone unnoticed during routine human patrols.
The research team, which includes Xue Zhongxin, Fan Shengjun, and Zhao Hongju from the China Coal Technology and Engineering Group Shenyang Research Institute and the State Key Laboratory of Coal Mine Safety Technology, emphasizes that this is not just a technological upgrade but a strategic shift toward “unmanned operation with remote supervision.” Their vision aligns with China’s national push for “reducing personnel and increasing efficiency” in the coal industry, a policy aimed at improving safety and competitiveness through automation.
The success of the Zhangjiamao system has broader implications for the global mining sector. While similar robotic systems have been tested in other countries, few have achieved such a high degree of integration in real-world, high-risk environments. The choice of a cable-dragged design over free-roaming autonomy may seem conservative, but it reflects a pragmatic understanding of the limitations of current technology in underground settings. By prioritizing reliability, safety, and regulatory compliance over novelty, the Chinese team has delivered a solution that is both innovative and deployable.
Experts in industrial automation note that the real breakthrough lies in the system’s holistic design. Rather than focusing on a single component—such as better sensors or faster processors—the team addressed the entire operational lifecycle: movement, power, communication, data processing, and human interaction. This systems-level approach is increasingly recognized as essential for the successful adoption of robotics in complex industrial environments.
Looking ahead, the team is exploring upgrades that could further enhance the robot’s capabilities. Potential additions include thermal imaging for detecting overheating equipment, acoustic sensors for identifying mechanical faults, and even robotic arms for simple maintenance tasks. They are also investigating the use of edge computing to perform more data analysis on the robot itself, reducing bandwidth requirements and enabling faster response times.
The deployment at Zhangjiamao is part of a larger trend toward intelligent mining, where artificial intelligence, the Internet of Things, and robotics converge to create safer, more efficient operations. As global demand for energy continues to evolve, and as the mining industry faces increasing pressure to improve safety and reduce environmental impact, solutions like this robotic inspection system offer a clear path forward.
In an industry where tradition often outweighs innovation, the Zhangjiamao project stands as a testament to what can be achieved when engineering rigor meets operational necessity. It is not just a robot that patrols a tunnel; it is a symbol of a new era in mining—one where human lives are protected not by luck or courage, but by intelligent design.
Mao Hao, Xue Zhongxin, Fan Shengjun, Zhao Hongju. Intelligent Inspection Robot System for Return Air Roadway in Zhangjiamao Coal Mine. Safety in Coal Mines. DOI: 10.13347/j.cnki.mkaq.2021.07.019