IMAGE PROCESSING BASED AUTONOMOUS LANDING ZONE DETECTION FOR A MULTI-ROTOR DRONE IN EMERGENCY SITUATIONS

Flight safety and reliability improvement is an important research issue in aerial applications. Multi-rotor drones are vulnerable to motor failures leading potentially unsafe operations or collisions. Therefore, researchers working on autonomous landing systems safely recover land the faulty drone a desired area. In such case, suitable zone should be detected rapidly for emergency landing. Majority of works related with utilize marker GPS signals detect site. this work, we propose system framework that involves only processing images taken from onboard camera vehicle. First, objects image determined by filtering edge detection algorithm, then most searched. The area free obstacles closest center defined as immediate zone. method has been tested 25 different heights its performance evaluated terms runtime single board computer precision recall values. average measured 2.4923 seconds 100% values achieved 1m 2m. smallest 79.1% 81.2%, respectively.