详情

原文传递 UAV Payload Identification With Acoustic Emissions and Cell Phone Devices.

题名：	UAV Payload Identification With Acoustic Emissions and Cell Phone Devices.
作者：	Doster, H. G.
摘要：	This thesis explores the ability to use cell phone devices for Unmanned AerialVehicle (UAV) acoustic payload detection. Past researchers demonstrated the abilityto use UAV acoustic signatures to determine whether a UAV carries a payload andthe weight of that payload. The experiments in past research were conducted at closerange with high-quality microphones. This research expands the field of study bytesting acoustic payload detection using cell phone devices and at far range. TheDepartment of Defense (DoD) is particularly interested in acoustic payload detectiondue to rising security threats from UAVs. Although these detection systems increasesecurity, the ever on-going arms race leads to counter techniques which make thesedetection systems ineffective. Therefore, there is a need for new drone detectionsystems which use new technology. Acoustic emissions are a unique property alldrones expel, and these emissions provide new stimulus for drone detection systems.An acoustic drone detection system does not require line-of-sight and is difficult tospoof, so an acoustic drone detection system which identifies payload weight usingcell phones would prove useful.Cell phones are commonplace worldwide. Due to this fact, there is a growingdesire to use cell phones for hazard detection. The ability to use a common cellphone to detect a far off hazardous UAV would improve security in many contestedenvironments.This research develops the prototype HurtzHunter to demonstrate acoustic pay-load detection with cell phone devices to collect UAV acoustic emissions, then usesthe emissions to train an AI capable of UAV payload classification. The HurtzHuntersystem tests acoustic payload detection with 7 different recording devices at 7 m, 10m, 20 m, 30 m, 40 m, 50 m, 75 m, and 100 m ground distance from the drone. Ateach distance, the experiment runs 6 flights each with a unique payload attached tothe drone. 80% of the acoustic emissions train a Support Vector Machine (SVM),and 20% tests the SVM.The methodology in this research shows the HurtzHunter design achieves an 82.81- 99.93% payload prediction accuracy based on the configuration. In short, thisresearch provides novel insight into the maximum range for UAV payload detectionusing acoustic emissions, and provides insight into the ability to use cell phone devicesfor payload detection.
总页数：	174 pages