Pseudo-range estimation for autonomous aircraft detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing passive range detection methods for autonomous vehicles, particularly light aircraft, are inadequate due to reliance on feature detection, multiple sensors, and complex processing, leading to delayed and inaccurate range estimates, which are not suitable for timely decision-making in detect and avoid scenarios.
Innovation Solution
A pseudo-range estimation method using a single passive sensor that combines detection data with prior performance models to generate a posterior probability distribution of pseudo-range estimates, allowing for immediate and rough range estimation without requiring feature detection or classification, and can be updated using Bayesian techniques and recursive filters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If model-based approaches are used for passive range detection, then instantaneous range estimates can be provided, but discernable features are typically not resolvable until well after initial detection, leading to delayed accurate range estimates
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing sensor performance models that characterize detection likelihood as a function of range under various conditions. These pre-computed models enable instantaneous range estimation at detection time without requiring feature resolution, resolving the contradiction between speed and precision by preparing the estimation framework in advance.
Solution Approach 2:
The patent changes the parameter being measured from feature-based characteristics (which require resolution time) to detection likelihood parameters based on sensor performance models. By transforming the estimation basis from visual features to sensor response characteristics, the system achieves both instantaneous estimation and accurate precision.
2Measurement precision
If higher resolution sensors are used for range measurement at appreciable distance, then measurement precision is improved, but device complexity, size, weight, and power demands increase
Solution Approach 1:
The patent substitutes mechanical/optical resolution improvements with a computational approach using sensor performance models. Instead of upgrading to higher resolution sensors that increase complexity and power demands, the system uses software-based probabilistic modeling to achieve accurate range estimation with existing sensor capabilities.
Solution Approach 2:
The patent changes the measurement parameter from direct spatial resolution to probabilistic detection likelihood. By estimating range through the relationship between detection likelihood and sensor performance models rather than through direct feature resolution, the system achieves precise range measurement without requiring complex high-resolution sensor systems.
3Adaptability or versatility
If multiple sensors are installed to cover minimum effective field of regard, then measurement coverage is improved, but weight, drag, and power consumption increase
Solution Approach 1:
The patent makes the sensor performance models universal by pre-computing detection likelihood characteristics that can be applied across different ranges and conditions. This single set of models serves multiple functions: estimating range, assessing detection confidence, and supporting decision-making, eliminating the need for multiple specialized sensors to achieve comprehensive coverage.
Solution Approach 2:
The patent creates a computational copy of sensor performance characteristics through pre-computed models rather than physically replicating sensors. The performance models serve as virtual representations of sensor capabilities across the entire field of regard, providing multi-functional coverage without the physical weight and complexity of multiple actual sensors.
4Measurement precision
If feature detection and classification are performed before range estimation, then measurement precision is improved, but loss of time occurs due to the sequential processing requirements
Solution Approach 1:
The patent applies preliminary action by pre-computing sensor performance models that encode the relationship between detection likelihood and range. This preliminary preparation eliminates the need for sequential feature detection and classification, as the models directly provide range estimates from detection data, achieving both precision and speed simultaneously.
Solution Approach 2:
The patent introduces sensor performance models as an intermediary that directly connects detection data to range estimates without requiring feature detection or classification as intermediate steps. This intermediary framework enables parallel processing of range estimation and feature analysis, eliminating sequential processing delays while maintaining accuracy.
Data Source
AI summary
A rough initial estimate of Line of Sight range (“pseudo-range”) is generated essentially immediately following the detection of an object by a passive sensor on a vehicle. The data are combined with prior detection likelihood and prior performance models for the sensor. These comparisons generate a posterior probability distribution of pseudo-range estimates. A pseudo-range estimate is derived from the probability distribution and output for use in detect and avoid decision-making and action planning. The pseudo-range estimate can be updated to improve its accuracy, such as by using a recursive filter (e.g., a Kalman filter). Other information, such as current atmospheric data, or known (or likely) vehicular activity in the region and at the current time, can be used in addition to the vehicle's spatial and temporal location, to improve accuracy.