Autonomous Landing Neural Network Aligning Flight Path Vector
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vision-based landing systems for aircraft rely on human pilots to interpret visual cues from head-up displays (HUDs), and there is no equivalent solution for autonomous systems to autonomously manage flight controls based on these cues.
Innovation Solution
An on-board computer system utilizing two neural networks to process video streams, runway data, and aircraft state data to identify landmarks and obstacles, and to determine flight control inputs necessary to align the flight path vector with a flight director, enabling autonomous landing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If vision-based landing systems use HUD visual cues with human pilots, then landing guidance and control are achieved, but the system cannot be applied to autonomous aircraft without human operators
Solution Approach 1:
The patent replaces the human pilot's visual interpretation and manual control operations with an autonomous system that uses computer vision algorithms and neural networks to process HUD visual cues and generate flight control commands, enabling autonomous aircraft to perform landing operations without human operators
Solution Approach 2:
The patent creates a digital representation of the visual scene by capturing images through onboard cameras and processing them through neural networks to identify runway landmarks and environmental features, allowing the autonomous system to interpret visual cues in the same way a human pilot would without requiring actual human presence
2Measurement precision
If the system processes video streams and identifies environmental landmarks, then accurate position detection is achieved, but computational complexity increases
Solution Approach 1:
The patent divides the complex task of visual scene interpretation into separate neural network modules: one network identifies runway landmarks and environmental features, while another processes flight path vector and flight director cues, allowing each module to specialize in specific functions and improving overall system efficiency
Solution Approach 2:
The patent employs neural networks that can process multiple types of visual information simultaneously - identifying runway thresholds, centerlines, touchdown zones, and environmental landmarks from the same video stream, making the system versatile for various landing conditions and runway configurations
3Manufacturing precision
If flight control inputs are determined to align flight path vector with flight director, then precise trajectory control is achieved, but control system complexity increases
Solution Approach 1:
The patent implements a feedback mechanism where the autonomous system continuously monitors the flight path vector and flight director cues, compares the current trajectory with the desired approach path, and adjusts flight control inputs in real-time to maintain alignment, ensuring precise landing trajectory control
Solution Approach 2:
The autonomous system independently determines the necessary flight control inputs by processing visual cues and comparing them with desired flight parameters, without requiring external pilot intervention, allowing the aircraft to self-correct its trajectory and complete the landing autonomously
Data Source
AI summary
An on-board computer system includes a first neural network trained to receive a video stream of a landing approach, runway data, and aircraft state data and identify environmental landmarks corresponding to components of the runway and environmental obstacles that might interfere with autonomous landing. The on-board computer system also includes a second neural network that receives a flight path vector and a flight director corresponding to vectors based on the aircraft energy state and a desired aircraft flight path respectively. The second neural network determines flight control inputs to bring the flight path vector into conformity with the flight director.