An Accurate Line Detection by a Novel Line Follower Sensor Output Processing Method

Line sensors are widely used in line following robots. Their performances are at a premium in the line following task of Automated Guided Vehicles (AGV) at hobby or industrial scale. In this manuscript a a line following sensor (LFS) array output processing method that significantly enhancing the performance of the LFS array is presented. This processing method makes use of cubic Bezier curves and extracts in a highly accurate position information from the LFS outputs. A line sensor array configuration under consideration consists of M sets of infrared (IR) emitting LED and IR sensitive photodetector, called an L-Pd set. The sets are equispaced and collinearly positioned. In the mode of operation light from the IR emitting LEDs is reflected back onto the photosensitive detectors. Depending on the intensity of the light received, each photodetector outputs an analog voltage between Vmin to Vmax volts. Because the L-Pd set components are possibly not identical, each one’s voltage response to all-black, respectively all-white, targets are not equal. Normalization of the L-Pd sets is achieved by the aid of linear parametrizing. Having the normalized L-Pd sets the remaining problem can be stated as detecting the center of a black line strip placed on a white background by analyzing the voltage outputs of the L-Pd sets in the line sensor array. At this stage of the problem the voltage outputs of the L-Pd sets versus the sensor locations is formed, and transformed into a sequence cubic Bezier curves. An efficient algorithm presented in this manuscript yields the center of the black strip. An experimental set–up presented in this manuscript shows validity and accuracy of the approach outlined above.