Automated Analogue Dial Reading in Cockpits

A number of functional aircrafts still feature traditional cockpits with analogue gauges and dials in the modern world. In order to simplify aircraft navigation and allow the pilot to focus only on the most important information, technological intervention is imperative. This would also help minimize human error, log data and keep up with advancements in the modernization of aircraft displays. This paper has been focused on automating the task of analogue dial reading in cockpits, when there is a fixed-position camera available for capturing cockpit panel images. The Single Shot Detector Mobile Net v2(SSD) is used for gauge detection and cropping as it brings a good balance between speed and reliability. The Automated Analogue Dial Reading (AADR) system then preprocesses the detected gauges and finds candidate pointer lines that constitute the needle entity, using the Hough transform method. An optimal line depicting the pointer of the needle is then found from the lot, and the reading of the dial is obtained by associating the inclination of the needle with the dial configurations. Experimental results reveal that the cockpit gauges are detected from cockpit images with an accuracy of 98.64%, and the readings obtained are 98% accurate.