Two-Stage Vision Guidance for Long-Range UAV Landing Accuracy

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current autonomous UAV landing technologies are limited by small visual recognition angles and short recognition distances, restricting the accuracy and efficiency of UAVs landing on unmanned vehicle platforms, especially in agricultural applications where precise and long-range guidance is necessary.

Innovation Solution

A vision-based method using a YOLOv5 neural network to detect and recognize unmanned vehicles, employing a target pattern with AprilTags and an H-shaped geometric pattern, enabling long-range and short-range guidance for precise UAV landing by collecting and processing images to adjust flight status and altitude.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional visual recognition markers (large QR code, equilateral triangle with concentric circles) are used for UAV landing guidance, then the UAV can identify the landing area and perform attitude adjustment, but the viewing angle is limited and recognition distance is short

Engineering Contradiction:
Improvelanding accuracyVSAvoidrecognition distance
Core Design Contradiction:
Measurement precisionVSLength of stationary object

Solution Approach 1:

The patent transitions from 2D planar markers (QR codes, triangles) to a 3D vertical structure (H-shaped geometric pattern with AprilTags at different heights). This dimensional change allows the UAV to recognize the target from a broader range of distances and angles, effectively extending the recognition distance while maintaining landing precision.

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

Solution Approach 2:

The patent embeds multiple recognition elements within a single structure: AprilTags are nested within the H-shaped geometric pattern. This nested design allows the system to first detect the large H-shape for long-range guidance, then use the embedded AprilTags for short-range precise positioning, thereby extending effective recognition distance across multiple scales.

Inventive Principle:
Principle #7Nested doll (Nesting)

2Measurement precision

If only close-range target recognition is used, then the UAV can achieve precise positioning, but the effective landing guidance distance is limited

Engineering Contradiction:
Improvepositioning accuracyVSAvoidtime to reach landing position
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary long-range detection using the H-shaped geometric pattern visible from greater distances. This allows the UAV to identify the landing area and begin navigation earlier, reducing the time to reach the landing position. The embedded AprilTags then provide preliminary fine-tuning information as the UAV approaches, further optimizing the overall timing.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The landing guidance process is segmented into two phases: long-range guidance using the H-shaped pattern for coarse positioning, and short-range guidance using AprilTags for fine positioning. This segmentation allows each stage to optimize for its specific distance range, improving overall efficiency and reducing total time to landing.

Inventive Principle:
Principle #1Segmentation

3Device complexity

If traditional fixed targets are used for landing guidance, then the system is simple to implement, but it cannot effectively control autonomous landing onto moving unmanned vehicle platforms

Engineering Contradiction:
Improvesystem complexityVSAvoidadaptability to moving platforms
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent transitions from static fixed targets to a dynamic target system that moves with the unmanned vehicle platform. The H-shaped geometric pattern and AprilTags are mounted on the moving platform, allowing the UAV to track and adapt to the platform's motion in real-time, thereby achieving effective autonomous landing on moving targets without significantly increasing system complexity.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS20240411316A1Method of vision-based long-range and short-range guidance for autonomous UAV landing
Publication Date: 2024.12.12 SHANGHAI UNIV
  • US20240411316A1 patent drawing
  • US20240411316A1 patent drawing
  • US20240411316A1 patent drawing

AI summary

Provided is a method for autonomous unmanned aerial vehicle (UAV) landing, including: collecting an unmanned vehicle image dataset in advance, training an unmanned vehicle detection model by using the unmanned vehicle image dataset combined with a YOLOv5 neural network; collecting, by the UAV during a landing process, images of an area below the UAV at specified time intervals, inputting the collected images into the unmanned vehicle detection model for recognition and detection; if an unmanned vehicle is recognized, further determining position information of the unmanned vehicle, and outputting, by a control module, a long-range guidance control instruction, to instruct the UAV to fly to a specified distance position above the unmanned vehicle; collecting, by the UAV, an image of a target and determining position information of the target, and outputting, by the control module, a short-range guidance control instruction to instruct the UAV to land on an unmanned vehicle platform.