Autonomous Collision Avoidance and Mapping for a Quadrotor Using Panning Sonar Sensors

Recent research and development of autonomous vehicles has expanded their capabilities in both military and civilian applications. A variety of missions including military and police surveillance, scientific reconnaissance, and search and rescue in large or hazardous search areas can now be done more efficiently and effectively through the use of autonomous vehicles. In order to decrease workload and reduce risk of further human life, the research here focuses on implementation of an autonomous navigation algorithm to allow a quadrotor to successfully maneuver through its environment. Using two ultrasonic sonar sensors and implementing a Vector Field Histogram (VFH) algorithm in conjunction with an A* algorithm, a technique is created that demonstrates successful navigation through a dense environment to a prespecified waypoint without any prior environmental knowledge. Furthermore, using a voting and de-voting scheme on a Cartesian grid provides robustness to spurious or inconsistent sensor data. The algorithm is tested in a simulation for a quadrotor.