December 23, 2024
While commercial UAV news primarily focuses on real-world use cases that demonstrate how drones can help businesses and organizations achieve their goals, we also provide a variety of It is important to remember that there are many researchers working “behind the scenes.” .
Commercial UAV News recently had the opportunity to speak with one of these researchers. krishna muvvaa doctoral candidate at the University of Nebraska-Lincoln, researching aeronautical robotics.
“My work involves artificial intelligence, including robotics, computer vision, and machine learning,” he explained. “And we have been researching ways to integrate MPC and neural network-based approaches to help drones achieve precise navigation, obstacle avoidance, and more robust control in complex environments.”
Through our work, Muvva aims to: Using a combination of traditional and AI-based control methods and advanced perception mechanisms, “UAVs can move and operate autonomously in dynamic environments such as public safety, critical infrastructure, and agriculture. Find a way to do it.
for example:
The paper, “Cooperative localization of UAVs in multi-robot systems using deep learning-based detection,” co-authored by Muvva, addresses the question of how UAVs can achieve accurate localization, especially when flying in close proximity. I’m working on it. In this paper, Muvva and his colleagues present a new framework that “leverages deep learning-based detection” to improve localization accuracy for multi-robot systems consisting of UAVs and unmanned ground vehicles (UGVs). I’m explaining how I came up with it. Through this approach, researchers are able to improve “localization accuracy even in difficult environments,” which could enable improved drone operations in fields such as agriculture, law enforcement, and mapping. There is sex.
The video below demonstrates cooperative localization, where multiple UAVs work together to enhance location awareness using deep learning-based detection and communication.
Aiming to further integrate advanced technology into drone operations, Muvva and his colleagues have been aiming to improve the awareness of convolutional neural networks (CNNs) for autonomous systems like drones. As described in the study “Reliable AI for UAVs through Control/Perception Co-Design,” researchers focused on the challenges posed by the trade-off between delay and accuracy in CNNs. They developed a “situational awareness model predictive control (MPC) for integrating perception and control modules.” Through both real-world and simulator testing, this approach has been shown to help drones achieve optimal system performance.
For Muvva, these studies pave the way to strengthen the integration of AI into drone missions, potentially leading to more efficient, accurate, and effective operations.
““As UAV applications expand into complex environments, achieving accurate navigation, obstacle avoidance, and robust control becomes more important,” he said. “Applying hybrid MPC with neural networks for UAV adaptability in difficult environments demonstrates the transformative potential of AI in UAV operations.”
