Indoor fire accidents have become increasingly common in recent years. More and more firefighting robots have been designed to solve the problem of fires. However, the indoor environment is very complex, with high temperatures, thick smoke, more turns, and various burning substances. In this study, a firefighting robot with autonomous inspection and automatic fire-extinguishing functions intended for use in unknown indoor environments was designed. Considering water’s poor efficiency and its inability to extinguish some combustion materials, other fire extinguishers were applied to design the extinguishing system. The robot can install four different extinguishers as required and select the appropriate fire extinguisher to spray it automatically. Based on the Cartographer SLAM (simultaneous localization and mapping) theory, a robot map-building system was built using Lidar scanners, IMU (inertial measurement unit) sensors, encoders, and other sensors. The accurate identification and location of the fire source were achieved using an infrared thermal imager and the YOLOv4 deep learning algorithm. Finally, the performance of the firefighting robot was evaluated by creating a simulated-fire experimental environment. In an autonomous inspection process of the on-fire environment, the firefighting robot could identify the flame in real-time, trigger the fire-extinguishing system to carry out automatic fire extinguishing, and contain the fire in its embryonic stage.
Fire robots are an effective way to save lives from fire, but their limited detection accuracy has greatly hampered their practical applications in complicated fire conditions. This study therefore proposes an advanced thermal imaging flame detection model of YOLOv4-F based on YOLOv4-tiny. We replaced the Leaky ReLU activation function with the Mish activation function in the YOLOV4-tiny feature extraction network. A Spatial Pyramid Pooling (SPP) was also added to increase the receiving range of the feature extraction network. To improve the feature fusion efficiency between multi-scale feature layers, a Path Aggregation Network (PANet) was adopted to replace the YOLOv4-tiny Feature Pyramid Network (FPN) with full use of feature information; a high-quality dataset containing 14,757 thermal imaging flame images was built according to the PASCAL VOC 2007 dataset standard. The results show that, when compared to the YOLOv4-tiny, YOLOv5-s, and YOLOv7-tiny models, the average detection accuracy of the proposed YOLOv4-F model is 5.75% higher, the average mAP of the five IOU cases rises by 7.02%, and the average detection confidence of three scaled flames shows a 18.09% gain. The proposed YOLOV4-F meets the requirements of fire robots on real-time responses and accurate flame detection, offering an important tool to improve the performance of the current fire robots.
Light detection and ranging (LiDAR) is one of the popular technologies to acquire critical information for building information modelling. To allow an automatic acquirement of building information, the first and most important step of LiDAR technology is to accurately determine the important gesture information that micro electromechanical (MEMS) based inertial measurement unit (IMU) sensors can provide from the moving robot. However, during the practical building mapping, serious errors may happen due to the inappropriate installation of a MEMS-IMU. Through this study, we analyzed the different systematic errors, such as biases, scale errors, and axial installation deviation, that happened during the building mapping, based on a robot equipped with MEMS-IMU. Based on this, an error calibration model was developed. The problems of the deviation between the calibrated and horizontal planes were solved by a new sampling method. For this method, the calibrated plane was rotated twice; the gravity acceleration of the six sides of the MEMS-IMU was also calibrated by the practical values, and the whole calibration process was completed after solving developed model based on the least-squares method. Finally, the building mapping was then calibrated based on the error calibration model, and also the Gmapping algorithm. It was indicated from the experiments that the proposed model is useful for the error calibration, which can increase the prediction accuracy of yaw by 1–2° based on MEMS-IMU; the mapping results are more accurate when compared to the previous methods. The research outcomes can provide a practical basis for the construction of the building information modelling model.
The terahertz imaging detection technology is applied to low-visibility fire environments to improve the traditional imaging detection equipment affected by the changes in the morphology and particle size distribution of smoke particles. This study focus on the transmission characteristics of terahertz waves in fire and smoke environments. A model for the propagation of terahertz waves is obtained. Terahertz imaging equipment is used to experimentally investigate the flue gas environment generated by three standard fires and combined with numerical simulations to analyze the influence of the terahertz wave transmission under different flue gas environments. The results show that the transmission of terahertz waves in a flue gas environment is affected by factors such as the smoke particle size and flue gas concentration; however, the transmission is significantly better than that obtained with traditional detection methods. In standard fire experiments, the terahertz equipment can achieve good imaging results, and the transmittance of terahertz waves is maintained above 88%.
To address the current issues of path planning and limited applicability in patrol robots, an enhancement has been made to the existing A* algorithm, coupled with the utilization of the dynamic window approach, to design a global dynamic path planning method suitable for patrol robots. This approach involves introducing dynamic weights to enhance search efficiency, identifying collinearity among global path nodes to eliminate redundant points, and conducting obstacle detection between essential nodes to enhance path smoothness. An autonomous navigation algorithm is proposed, wherein the path generated by the A* algorithm is adopted as the global path. The essential nodes within this path are treated as local target points. By adjusting the initial pose angle and optimizing the weight coefficients of the evaluation function, the autonomous navigation algorithm’s optimization efficiency and obstacle avoidance capability against random obstacles are improved. Comparative simulation experiments are conducted to validate its robust performance in terms of planning efficiency, path length, and random obstacle avoidance. By establishing a ROS robot operating platform and conducting experiments in both static and dynamic environments, the autonomous navigation algorithm is shown to successfully plan superior paths and achieve real-time obstacle avoidance to reach the target point.