Aircraft Rerouting With ML-Assisted Shortest-Path Finding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current aircraft path re-routers based on rule-based shortest path finding (SPF) algorithms face limitations when handling multiple parameters, leading to increased latency and difficulty in incorporating accelerating weather and pilot intuition, especially in real-time re-routing scenarios.
Innovation Solution
Integration of machine learning (ML) with SPF algorithms to infer optimal re-route paths by considering various parameters, using a trained ML model to reduce the load on SPF algorithms and enhance decision-making in real-time re-routing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If rule-based SPF algorithms are used for path re-routing, then the system can find a flight path in the presence of hazards, but the processing load increases and latency increases when considering forty or more parameters
Solution Approach 1:
A trained machine learning model is introduced as an intermediary between the input parameters and the SPF algorithm. The ML model processes the forty or more parameters and outputs a reduced set of relevant parameters and initial path suggestions, which then feed into the SPF algorithm. This intermediary layer reduces the computational burden on the SPF algorithm while maintaining path finding accuracy.
Solution Approach 2:
The path re-routing process is segmented into two distinct stages: (1) ML-based parameter processing and initial path suggestion, and (2) SPF-based optimization and final path determination. This segmentation allows each component to specialize in specific tasks, with the ML model handling high-dimensional parameter reduction and the SPF algorithm focusing on precise path optimization, thereby reducing overall latency.
2Reliability
If rule-based SPF algorithms are used for path re-routing, then the system can handle hazards on the flight path, but the system has difficulty incorporating accelerating weather and pilot intuition
Solution Approach 1:
The system transforms qualitative pilot intuition and weather risk assessments into quantifiable parameters that the ML model can process. Pilot best practices and weather forecast risk perceptions are converted into numerical inputs that influence the ML model's predictions, enabling the system to adapt to accelerating weather conditions and incorporate pilot expertise dynamically.
Solution Approach 2:
The machine learning model continuously learns from new data including pilot behaviors and weather patterns, automatically updating its internal representations without requiring explicit reprogramming. This self-learning capability allows the system to adapt to changing conditions and incorporate pilot intuition over time, enhancing versatility while maintaining hazard avoidance reliability.
Data Source
AI summary
A system may include a processor configured to: (a) obtain parameters; (b) based on the parameters, update flight-state data associated with an aircraft; (c) obtain a trained machine learning (ML) model; (d) based at least on the updated flight-state data and the trained ML model, infer a direction from a current cell for a reroute; (e) based on the inferred direction and the updated flight-state data, set the current cell and identify neighboring cells; (f) calculate an optimal next cell by using a shortest path finding (SPF) algorithm to select the optimal next cell from the neighboring cells; (g) iteratively repeat steps (d) through (f) such that the current cell is set as the optimal next cell until a goal state is reached; (h) construct a re-route using optimal cells iteratively calculated in step (f); and (i) output the re-route.