Incremental Neural Network Inference for Spacecraft Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current space navigation systems rely on ground-based control methods, which are inefficient due to time lags in communication, leading to outdated navigation data and increased uncertainty in spacecraft state.
Innovation Solution
An incremental neural network model inference method is implemented onboard the spacecraft, allowing for real-time updates of navigation state estimates using a neural network model with reduced memory requirements, enabling immediate course corrections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If ground-based control methods are used for space navigation, then system complexity is reduced, but navigation update time increases and accuracy decreases
Solution Approach 1:
The patent segments the navigation computation process into discrete incremental steps that can be executed sequentially on onboard hardware. Instead of requiring complete recalculations, the system processes navigation updates in manageable segments that reduce computational burden while maintaining real-time performance.
Solution Approach 2:
The patent implements preliminary action by pre-computing and storing navigation parameters and algorithms on the onboard computer before missions begin. This allows the spacecraft to perform rapid incremental updates during flight without requiring complex real-time calculations, effectively preparing the system in advance for efficient operation.
2Speed
If complete neural network models are loaded into memory, then computation speed improves, but memory requirements increase
Solution Approach 1:
The patent divides the neural network model into segmentable components that can be loaded and processed in parts rather than requiring the entire model to reside in memory simultaneously. This segmentation allows efficient computation with reduced memory footprint by processing different layers or subsets of the network sequentially.
Solution Approach 2:
The patent implements partial action by loading only the necessary portions of the neural network model into memory at any given time rather than the complete model. This approach maintains adequate computation speed for navigation tasks while significantly reducing the total memory capacity required on the spacecraft.
Data Source
AI summary
An incremental neural network model inference method for updating a navigation state estimate of a spacecraft includes: reading parameters and functions from a current layer of a neural network model into a memory of a spacecraft computer, performing a transformation on a navigation state estimate with the parameters and functions, saving a transformed navigation state estimate to the memory, and determining whether the current layer of the neural network model is the last layer. When the current layer is not the last layer, the method includes: removing the parameters and functions from the memory; incrementing to a next layer of the neural network model; repeating the steps of reading, transforming, and removing for the next layer of the neural network model until all layers are transformed; when the current layer is the last layer, the method includes: outputting a thrust command.