Autonomous UAS Terrain Mapping with Coarse-to-Fine Landing Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing unmanned aerial vehicles (UAVs) face challenges in autonomously detecting safe landing sites in complex 3D terrain due to limited computational resources and the inability to accurately perceive and understand their surroundings, especially in GPS-denied environments, where existing sensors and algorithms are either too heavy, power-consuming, or computationally demanding.
Innovation Solution
A lightweight and energy-efficient system for UAVs that uses a Structure-from-Motion algorithm coupled with an on-board pose estimator to generate dense depth maps, which are then fused into a robot-centric, multi-resolution digital elevation map, employing Kalman and Optimal Mixture of Gaussian filters to update and maintain a consistent environment model, and apply coarse-to-fine search based on slope, roughness, and uncertainty for safe landing site detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing sensors and algorithms are used for terrain mapping and landing site detection, then measurement precision and reliability are improved, but device complexity, weight, and energy consumption increase
Solution Approach 1:
The patent segments the terrain mapping process into multiple resolution levels (coarse and fine). The coarse-to-fine search strategy divides the detection space into hierarchical levels, where coarse-level processing identifies general safe zones and fine-level processing refines landing site selection. This segmentation reduces computational complexity while maintaining detection precision by applying appropriate processing intensity only where needed.
Solution Approach 2:
The patent applies partial action by using simplified algorithms for coarse terrain assessment and reserved excessive computational resources only for fine-detail processing in identified safe zones. Instead of applying full-resolution processing uniformly across all terrain, the system performs intensive computation only in regions that pass initial coarse filtering, thereby reducing overall computational burden while maintaining precision in critical areas.
2Measurement precision
If high-resolution terrain mapping is performed to detect safe landing sites, then measurement precision is improved, but use of energy and computational resources increases
Solution Approach 1:
The patent segments the energy consumption profile by resolution level. Low-resolution processing is applied to the entire terrain area for initial assessment, consuming minimal energy. High-resolution processing is applied only to small candidate regions identified at coarse levels, consuming significant energy only where necessary. This segmentation of computational effort by spatial and resolution domains dramatically reduces total energy consumption while maintaining detection accuracy.
Solution Approach 2:
The system performs partial high-resolution processing only in candidate landing zones rather than across the entire terrain. The coarse-to-fine strategy ensures that computationally intensive fine-detail analysis is applied partially only to regions that pass initial screening, avoiding excessive energy consumption in areas that will ultimately be rejected as unsuitable landing sites.
3Productivity
If real-time landing site detection is implemented in unknown terrain, then productivity and response time are improved, but device complexity and computational demand increase
Solution Approach 1:
The patent segments the detection process into sequential stages (coarse scanning followed by fine search) that can be executed in real-time. Each stage processes data at appropriate resolution levels and produces intermediate results that guide the next stage. This segmentation enables real-time processing by breaking down the complex task into manageable computational steps that can be executed within available time constraints on embedded systems.
Solution Approach 2:
The coarse-to-fine strategy performs preliminary action by conducting rapid coarse-level terrain assessment before committing to detailed analysis. This preliminary processing identifies candidate regions and eliminates unsuitable areas early, preparing the data structure and reducing the search space for subsequent fine-detail processing. This preliminary action enables real-time response by ensuring that intensive computation is always applied to already-filtered, promising candidates.
Data Source
AI summary
A method, system, and apparatus for an unmanned aerial vehicle (UAV) to autonomously reconstruct overflown terrain and detect safe landing sites. A UAV autonomously acquires on-board pose estimates from an on-board visual-inertial-range odometry method during flight. The on-board pose estimates are utilized as a pose prior and to regain metric scale during three-dimensional (3D) reconstruction. The on-board pose estimates are corrected based on a bundle adjustment approach using previously acquired images. 3D reconstruction is performed based on multiple captured images taken from an on-board camera. Range data from the multiple captured images is fused into a multi-resolution height map. A safe landing site on the terrain is detected based on the multi-resolution height map.