CarderPlanet
Professional
No tree will interfere with the tracking of flying zaslants any more.
A team of researchers from the University of Texas and the University of Tennessee in the United States has developed a DroneChase system based on Raspberry Pi, which allows you to continuously track the flight path of drones even if you lose visual contact. The system uses a combined approach, combining acoustic and visual signals, which allows it to function in conditions of limited visibility.
DroneChase consists of 6 microphones, a video camera and a low-power Raspberry Pi computer. The entire system is placed on a mobile platform, for example, on a car, which allows you to continuously track the drone, without being tied to a specific location.
During DroneChase training, the YOLOv5 visual model was used to automatically generate drone position labels on images. These tags were then passed to the acoustic model, which was trained to determine the location of the drone only by sound.
As a result, the DroneChase system is able to locate the drone with an accuracy of 61% from the audio signal, even if visual contact is lost, for example, when the drone is hidden behind trees or in the dark. This indicator significantly exceeds traditional visual tracking methods.
According to the developers, the system is mobile and affordable thanks to the use of Raspberry Pi. In the future, it is planned to expand the functionality for tracking multiple drones simultaneously.
Overall, DroneChase demonstrates the strong potential of combined acoustic and visual approaches to improve the safety of drone use.
A team of researchers from the University of Texas and the University of Tennessee in the United States has developed a DroneChase system based on Raspberry Pi, which allows you to continuously track the flight path of drones even if you lose visual contact. The system uses a combined approach, combining acoustic and visual signals, which allows it to function in conditions of limited visibility.
DroneChase consists of 6 microphones, a video camera and a low-power Raspberry Pi computer. The entire system is placed on a mobile platform, for example, on a car, which allows you to continuously track the drone, without being tied to a specific location.
During DroneChase training, the YOLOv5 visual model was used to automatically generate drone position labels on images. These tags were then passed to the acoustic model, which was trained to determine the location of the drone only by sound.
As a result, the DroneChase system is able to locate the drone with an accuracy of 61% from the audio signal, even if visual contact is lost, for example, when the drone is hidden behind trees or in the dark. This indicator significantly exceeds traditional visual tracking methods.
According to the developers, the system is mobile and affordable thanks to the use of Raspberry Pi. In the future, it is planned to expand the functionality for tracking multiple drones simultaneously.
Overall, DroneChase demonstrates the strong potential of combined acoustic and visual approaches to improve the safety of drone use.
