A Novel Method of UAV-Assisted Trajectory Localization for Forestry Environments.

Area of Science:

• Robotics and Autonomous Systems
• Geospatial Intelligence
• Wireless Communication

Background:

• Global Positioning Systems (GPS) are unreliable in dense forest canopies.
• Existing localization methods require fixed infrastructure, unsuitable for dynamic forestry.
• Environmental uncertainties degrade signal quality and reduce localization accuracy.

Purpose of the Study:

• To develop an innovative trajectory localization method for forestry environments.
• To address the limitations of traditional localization systems in complex terrains.
• To enhance localization accuracy and reliability using Unmanned Aerial Vehicles (UAVs).

Main Methods:

• Utilizing multi-agent deep reinforcement learning (DRL) to optimize UAV topology in real-time.
• Employing Received Signal Strength (RSS) measurements from UAVs to the target.
• Implementing a least squares algorithm for flexible and reliable location estimation.
• Incorporating shared replay memory to improve DRL system performance and efficiency.

Main Results:

• The proposed UAV-assisted method achieves flexible and high-accuracy trajectory localization.
• Demonstrated superior robustness against high-dimensional heterogeneous data compared to existing systems.
• Validated suitability for challenging forestry environments through simulations.

Conclusions:

• The novel multi-agent DRL approach offers a robust and accurate solution for localization in forests.
• UAVs, when strategically deployed, significantly enhance localization capabilities in GPS-denied areas.
• This method provides a reliable alternative for forestry applications requiring precise spatial awareness.