Energy-saving UAV object detection system based on edge computing

Unmanned aerial vehicles (UAVs), commonly referred to as drones, have been used in countless environments to solve real-world problems. These flying robotic systems can help monitor the natural environment, detect fires or other environmental hazards, monitor cities, and search for survivors of natural disasters.

To tackle all these tasks effectively, a UAV must be able to reliably detect targets and objects of interest in its surroundings. Therefore, computer scientists have been trying to create new computational techniques that can enable these capabilities using deep learning or other approaches.

Researchers at Yunnan University and the Chinese Academy of Sciences recently introduced a new object detection system based on edge computing. Their recommendation system, introduced in IEEE Internet of Things Magazinecan provide UAVs with the ability to detect related objects and targets in their surroundings without significantly increasing their energy consumption.

Jiashun Suo, Xingzhou Zhang, Weisong Shi and Wei Zhou wrote in their paper.

“We present E3-UAV, an edge-based energy-saving object detection system for UAVs. This system is designed to flexibly support UAV devices, edge devices, and generation algorithms. different performance, aiming to minimize energy consumption by deciding at most energy-saving flight parameters (including flight altitude, airspeed, detection algorithm and sampling rate) ) required to meet the mission’s detection requirements.”

The object detection system invented by this group of researchers, named E³-UAV, based on an increasingly popular method known as edge computing. Edge computing leverages multiple neighboring networks or devices to perform computations faster and with less power consumption. In the case of the team’s system, these networks are leveraged to determine parameters (i.e. UAV altitude, airspeed, etc.) that help the system detect objects in the surrounding environment while focusing consume as little energy as possible.

Suo, Zhang and colleagues write in the paper: “We first present a performance measure for actual missions and build a transparent energy consumption model based on hundreds of trip data. actual flight to formalize the relationship between energy consumption and flight parameters”. “We then present a lightweight energy-saving priority decision algorithm based on a large amount of actual flight data to assist the system in deciding flight parameters.”

Suo, Zhang, and colleagues trained and evaluated their system with a series of simulations running on NVIDIA GPUs. They specifically applied it to the Mavic Air 2, a drone created by DJI and commonly used for aerial photography and video.

Suo, Zhang and their colleagues write: “We evaluate the performance of the system, and our test results demonstrate that it can significantly reduce energy consumption in real-world scenarios.” . “In addition, we provide four insights that can assist researchers and engineers in their efforts to further investigate UAV-based object detection.”

In the future, E³-UAV can be deployed and tested on other UAVs to further assess its potential and generality. In addition, this work could inform the development of edge computing-based similarity object detection techniques for robotic applications.