Drone deliveries provide an innovative alternative to current last-mile delivery solutions; however, drones are limited by their short flight time. To overcome this, researchers devised the Multi-Mode Hybrid Drone Delivery System, in which a plane transports a drone to deliver a package. The preexisting system requires improvements to its computer vision algorithms to enable the drone to redock to the plane. This research aims to evaluate and improve the performance of the object detection algorithm. Three tests are performed to establish performance metrics: the Frequency, Distance Tracking, and Horizontal Position Accuracy Tests. Results reveal that decreasing the dimensions of the camera’s input significantly increases the processing speed of the computer vision algorithm. A new method is proposed that improves the accuracy of the drone’s positional tracking, providing the precision needed to make the redocking process possible.
Improving the Performance of Computer Vision Algorithms for the Multi-Mode Hybrid Drone Delivery System
30.09.2022
1407858 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
DELIVERY DRONE, DRONE PORT AND AUTOMATED DELIVERY SYSTEM
Europäisches Patentamt | 2022
|DELIVERY DRONE, DRONE PORT AND AUTOMATED DELIVERY SYSTEM
Europäisches Patentamt | 2022
|