Hierarchical Multi-Drone Imaging for Adaptive Scan Task Planning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for planning and controlling drone trajectories in a multiple drone swarm are inefficient, requiring lengthy preparation times and manual decision-making, especially when adapting to variations in spatial density of features in a scene, which limits the quality and efficiency of image capture and reconstruction.
Innovation Solution
A hierarchical drone system where a root drone captures keyframe images and generates ground mask images to direct level-1 and level-2 drones to focus on regions of high interest, using wavelet transforms and window-based methods to determine scanning tasks and trajectories, allowing for adaptive and automated decision-making during flight.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If formatted swarm flight with precise formation and coordinated trajectory planning is used, then high quality image capture is achieved, but preparation time becomes lengthy and manual decision making is required
Solution Approach 1:
The system enables drones to autonomously plan their own trajectories and perform self-assignment of scanning tasks based on real-time scene analysis, eliminating the need for lengthy manual trajectory planning and preparation while maintaining high image capture quality
Solution Approach 2:
The trajectory planning becomes adaptive and dynamic, allowing drones to adjust their paths in real-time based on detected features of interest and spatial density variations, rather than following rigid pre-planned trajectories, thus reducing preparation time while maintaining capture quality
2Adaptability or versatility
If drones fly along different predetermined trajectories to capture global multi-view, then diverse perspectives are obtained, but trajectory planning becomes complex and time-consuming
Solution Approach 1:
A root drone performs a preliminary survey flight to gather scene data and identify features of interest before other drones execute their scanning tasks, enabling simplified individual trajectories while maintaining overall view diversity through hierarchical coordination
Solution Approach 2:
The scene is divided into multiple regions of interest identified by the root drone, and different level-1 drones are assigned to scan specific regions independently, simplifying each drone's trajectory planning while collectively achieving comprehensive scene coverage
3Adaptability or versatility
If preliminary survey flight is conducted to gather data for trajectory planning, then adaptive trajectory planning is enabled, but total preparation time becomes even longer
Solution Approach 1:
The root drone performs a brief preliminary survey and automatically processes the gathered data to generate scanning tasks for other drones, enabling adaptive trajectory planning without requiring extended manual preparation time, as the system autonomously converts survey data into actionable flight paths
4Duration of action of moving object
If not every drone views every part of the target, then flight time is reduced, but reconstruction quality is significantly limited
Solution Approach 1:
Different drones focus on different regions of interest based on spatial density analysis, with each drone capturing high-quality images of its assigned region, ensuring that every part of the target is viewed by at least one drone while maintaining overall reconstruction quality without requiring every drone to view every part
Data Source
AI summary
A method for optimizing image capture of a scene by a swarm of drones including a root drone and first and second level-1 drones involves the root drone following a predetermined trajectory over the scene, capturing one or more root keyframe images, at a corresponding one or more root drone orientations and root drone-to-scene distances. For each root keyframe image: the root drone generates a ground mask image for that root keyframe image, and applies that ground mask image to the root keyframe image to generate a target image. The root drone then analyzes the target image to generate first and second scanning tasks for the first and second level-1 drones to capture a plurality of images of the scene at a level-1 drone-to-scene distance smaller than the root drone-to-scene distance; and the first and second level-1 drones carry out the first and second scanning tasks respectively.