IV. SYSTEM INTEGRATION AND PROCESS FLOW
The computer vision algorithm is run using Matlab. Atmega328
communicates with matlab through serial port. Interface
of microcontroller with various peripherals is depicted
in Fig. 10. The entire process flow adopted in the fire fighter
is illustrated in Fig. 7. The robot initially starts scanning the
test environment through the camera in steps of 30 (step
angle should be less than the camera lens angle to avoid the
clipping of certain scenes) and at each step a frame is extracted
from the video and is processed for fire detection. If fire is
detected then the robot starts aligning towards the fire using
pixel mapping through (7). Once it is aligned, it starts moving
towards the detected fire. The motion of the robot is monitored
continuously and if the robot is not on course towards the
fire, it is then realigned through the same pixel mapping
algorithm. The temperature is continuously monitored through
the Temperature sensor (LM35) which acts as a reference
ensuring the direction of motion of the robot. The range (i.e.,
the distance from which it detects fire) of the UVTron sensor
is low (3-5mts) and hence this property is exploited to confirm
the presence of fire near the robot. Once the UVTron sensor’s
output crosses a certain threshold (This threshold is calibrated
based on the range of the extinguishing system), the microcontroller
is interrupted and the robot halts. Micro-controller
aligns the arm towards fire and activates the water pump. The status of the fire is continuously monitored using computer
vision and the water pump is switched off, once the fire is
extinguished completely.