Evaluating the Effectiveness of Simulated Cyber-Physical Attacks on Autonomous Drones

This study investigates the effectiveness of cyberattacks on autonomous quadcopters. Using the Webots simulation environment and a digital model of the Crazyflie 2.1 drone, we analyze four types of attacks: sensor bias, sensor noise, sensor delay, and communication interference. The impact of each attack is evaluated using three key metrics: mission deviation, mission delay, and failure rate. Our results highlight the need for robust filtering mechanisms to counteract even minor disruptions, and identify which sensors benefit most from such protection.