UAV Landing Obstacle Detection Using 3D Image Reconstruction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Unmanned aerial vehicles (UAVs) face challenges in safely transitioning from horizontal to vertical flight modes, particularly in identifying obstacles in landing areas with small spatial extent or similar colors/textures to the surrounding environment, which can lead to hazards during landing operations.

Innovation Solution

The system captures multiple images of the landing area using onboard cameras, processes them to determine pixel disparities using optical flow or stereo matching algorithms, generating a 3D reconstruction and difference images to detect obstacles, allowing the UAV to abort or modify the landing operation if obstacles are present.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If visual imaging data is used to identify obstacles in landing areas, then obstacles with distinct colors or textures can be readily identified, but obstacles with small spatial extent or similar colors/textures to the landing area cannot be reliably detected

Engineering Contradiction:
Improveobstacle detection accuracyVSAvoidobstacle visibility in imaging data
Core Design Contradiction:
Measurement precisionVSDifficulty of detecting and measuring

Solution Approach 1:

The patent transitions from 2D visual imaging data to 3D spatial representation by capturing images at multiple altitudes and constructing a three-dimensional model of the landing area. This dimensional transformation enables the detection of obstacles with small spatial extent or similar appearance to the landing area, as their spatial position and height become distinguishable in the 3D model even when they are indistinguishable in 2D images.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Solution Approach 2:

The system performs preliminary obstacle detection by capturing images at a first altitude before final landing, constructs a 3D model, and identifies obstacles in advance. This preliminary action allows the UAV to assess landing safety and abort or modify the landing operation before committing to the final descent, preventing potential collisions with undetected obstacles.

Inventive Principle:
Principle #10Preliminary action

2Reliability

If the UAV descends to a designated landing area, then it can perform precise landing operations, but it risks colliding with undetected obstacles that pose substantial hazards

Engineering Contradiction:
Improvelanding safetyVSAvoidcollision risk with obstacles
Core Design Contradiction:
ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The system performs preliminary obstacle detection and landing area assessment by capturing images at a first altitude, constructing a 3D model, and identifying obstacles before the final landing descent. This preliminary action enables the UAV to abort or modify the landing operation if obstacles are detected, ensuring safe landing operations by eliminating collision risks from undetected obstacles.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses imaging devices to capture images of the landing area, processes these images to generate 3D models, and uses this feedback information to determine whether to proceed with or abort the landing operation. The feedback loop continues during descent with images captured at multiple altitudes, allowing real-time assessment and adjustment of the landing decision based on detected obstacles.

Inventive Principle:
Principle #23Feedback

3Reliability

If multiple images are captured and processed to generate 3D reconstruction and difference images, then obstacle detection reliability is improved, but system complexity and processing time increase

Engineering Contradiction:
Improveobstacle detection reliabilityVSAvoidimage processing system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The image processing is segmented into distinct stages: capturing images at multiple altitudes, generating 3D models from these images, constructing difference images by comparing 3D models, and finally determining obstacle presence. This segmentation allows the complex processing task to be divided into manageable steps, each with a specific function, reducing overall system complexity while maintaining high detection reliability.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3685307B1Object detection and avoidance for aerial vehicles
Publication Date: 2023.09.06 AMAZON TECH INC
  • EP3685307B1 patent drawingFigure 1a
  • EP3685307B1 patent drawingFigure 1b
  • EP3685307B1 patent drawingFigure 1c

AI summary

Aerial vehicles that are equipped with one or more imaging devices may detect obstacles that are small in size, or obstacles that feature colors or textures that are consistent with colors or textures of a landing area, using pairs of images captured by the imaging devices. Disparities between pixels corresponding to points of the landing area that appear within each of a pair of the images may be determined and used to generate a reconstruction of the landing area and a difference image. If either the reconstruction or the difference image indicates the presence of one or more obstacles, a landing operation at the landing area may be aborted or an alternate landing area for the aerial vehicle may be identified accordingly.