Research On The Automation Of Water Cannon For Urban Fire Fighting Vehicle

Urban firefighting vehicle is defined as the first vehicle driven to the scene of fire,which is required to have strong abilities for emergency rescue and fire suppression.In order to guarantee the sufficient space for rescue equipments,the volume of water tank is usually low.Meanwhile,the water cannon mounted on the vehicle is generally with high flow rate,causing a faster consumption of extinguishant.Thus,portable water cannon is more widely used.However,portable water cannon requires firefighters to carry it towards fire,lay water hoses for it and adjust its posture in a dangerous condition.Up to now,most common methods to operate the desk water cannon are manual control with human strength and remote control by firefighters,which demands a relevantly high skill for firefighters.An experimental plate form for image-based water cannon was built with techniques such as machine vision,profibus and posture control for the sake of high precision and reducing the difficulty to handle the desk water cannon.An embedded controller was adopted to collect data of inclinometer through CAN bus and adjust the angle of water cannon with close-loop control.A high-performance computer functions as host computer to process images captured through ethernet and calculate the desired posture for water cannon.A stereo vision system was built with a couple of GigE Vision based industrial cameras.Disparity map of the scene of fire was generated using SGBM algorithm on the basis of a precisely calibration of image acquisition system.Spatial coordinanates of selected target were calculated by reprojecting the disparity map into threedimensional space and then control the water cannon with the mathematical model of water jet.Binary image of water jet was extracted by means of motion detection using background subtraction with Mixture of Gaussions background model and threshold segmentation in HSV color space.Trajectory of water jet was collected based on the binary image and was used for locating its drop point.Angle increment was determined with the offset between drop point and selected target,which was the precondition for close-loop control of the proposed system.Rapid preliminary control of water cannon was conducted based on the proposed experimental system.Close-loop control and its precision was simulated with the proposed algorithm using previously collected image sequences.And real-time performance of the entire system was evaluated by recording the time consumption of each process.The result showed up a satisfying result in both precision and real-time performance.Finally,all those works done were summarized and prospects of research interests around automatic water cannon and related technologies for firefighting were done at the end of the thesis.There are 102 figures,14 tables and 92 references in this thesis.