drone-optimization

ML for Drone Optimization

Enabling Beyond Visual Line of Sight (BVLOS) operations with drones is key to completely harness the potential of drones A critical aspect of BVLOS drone applications is optimizing the drone path.

A key challenge in the route optimization is that dynamic nature of the wireless channel. Since the channel is impacted by a variety of factors such as weather, user traffic, geography, etc., it is impossible to build a standard channel model that can be used to compute the optimal route.

BIG PICTURE

Beyond Visual Line of Sight (BVLOS) operation is key to enable applications that require large-scale drone deployment

Leveraging the power of cutting-edge technologies, specifically reinforcement learning, we are revolutionizing the way drones navigate and interact with their wireless environments and to ensure (a) Faster and more reliable connectivity for real-time control and response, (b) Higher throughput to transmit high definition video and sensor data in real time. (c) Improved safety by avoiding no fly zone or crowded areas, (d) Enhanced battery life

DELART ML TECHNICAL TEAM

Dr. Vignesh Manohar

Sr. Antenna Engineer
CEO, Delart Canada

Dr. Craig Kleski

Lead ML engineer
PhD – Mathematics – Univ of Virginia

Dr. Riten Gupta

Wireless Engineer
PhD – Wireless Engineering – UMich. Ann Arbor

How do we do this?

With Reinforcement Learning, we can have the drone ‘learn’ the characteristics of wireless environment

Challenge: RL model needs to be trained offline before it can be implemented

Solution: Generating real-world RF data in a controlled lab/simulation environment that can take into account several factors that impacts a wireless channel.

Methodology

REINFORCEMENT LEARNING ALGORITHM

NEXT STEPS

Delart Private 5G Networks promise superior performance over technology like Wi-Fi but that advantage comes with much greater technical complexity. Allow the cellular networking experts at Delart to design, build and manage your network so you can concentrate on serving your end users.

Contact us at

info@delartech.com