Path planning method and device for unmanned aerial vehicles providing traffic services

Communication drones that plan flight paths by predicting demand points using EMA-weighted traffic heatmaps

This technology is a path-planning method and device that allows unmanned aerial vehicles (UAVs) providing traffic services to autonomously determine their flight paths by dividing an area of interest into cells and analyzing traffic demand. By applying an Exponential Moving Average (EMA) to traffic heatmaps across multiple time slots within a time window, it creates a single weighted heatmap, designating the point with the highest request volume as the global target and intermediate waypoints in that direction as local targets.

Existing technologies have struggled to enable multiple UAVs to calculate optimal paths independently within a limited time without assistance from a control center. They also faced limitations in adjusting paths in real-time to reflect future traffic changes or coordinating path distribution among multiple drones.

This technology predicts future demand by smoothing heatmaps so that closer time slots are weighted more heavily. The global area is limited to the range reachable by the drone within the time window. The local area is narrowed to points reachable within a single time slot that have a positive dot product with the global target direction vector. Local targets are selected by scoring cells based on request volume, remaining processing capacity, and a weight that increases as the remaining time limit decreases; requests exceeding capacity are excluded. When encountering other drones, the global area is redefined to include their reachable zones. Applicable to disaster relief base stations, traffic distribution at large events, and mobile edge computing, this allows drones to proactively move to high-demand areas to process data requests within deadlines.

‍

‍

Key Features:

• Receiving traffic information representing the traffic demand of an area of interest divided into multiple cells
• Generating a single weighted traffic heatmap by applying EMA to the traffic heatmaps of multiple time slots within a time window
• Determining the point with the highest traffic request volume among grid points in the global area as the global target using the weighted traffic heatmap
• Defining a local area based on the global target and determining a local target toward the global target based on cell request volumes and time limits

‍

This invention was developed with support from the Ministry of Science and ICT’s AI Convergence Innovation Talent Cultivation program.

‍