Perceptive AV Navigation Using Asset Feature Frames Without GPS
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles (AVs) face challenges in navigating accurately in GPS-denied environments, such as underground, indoors, or under bridges, where traditional GPS-based location methods are insufficient for civilian inspection applications requiring high precision.
Innovation Solution
A system and method for perceptively navigating AVs using onboard-generated information, such as image data, to determine a fixed local frame and asset feature frame, allowing the AV to generate motion control commands for precise navigation without relying on GPS, utilizing sensors like cameras and LIDAR for feature detection and localization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS-based location determination is used, then navigation is simple and reliable in open environments, but navigation accuracy deteriorates to approximately 2 meters in GPS-denied environments such as underground, indoors, or under bridges
Solution Approach 1:
The navigation system segments the positioning task into multiple independent components: GPS-based positioning for open environments, visual odometry for feature-based tracking, and inertial navigation for continuous pose estimation. Each component operates independently and can be activated based on environmental conditions, allowing the system to maintain high accuracy across diverse environments including GPS-denied areas.
Solution Approach 2:
The system implements a universal navigation framework that integrates multiple positioning methods (GPS, visual odometry, inertial navigation) into a single system. The perceptive navigation subsystem can universally handle both GPS-available and GPS-denied environments by switching between or combining these methods, making the navigation system adaptable to all operational conditions without requiring separate specialized systems.
2Measurement precision
If traditional GPS-based navigation is used, then the system is simple to implement, but it cannot achieve within 2 meters accuracy required for civilian inspection applications
Solution Approach 1:
The system introduces an intermediary perceptive navigation subsystem that acts as a mediator between the simple GPS receiver and the complex requirement for high-precision navigation. This subsystem processes image data from cameras, extracts environmental features, and computes positional poses relative to these features, thereby achieving meter-level accuracy without requiring complex hardware modifications while maintaining system architecture simplicity.
Solution Approach 2:
Instead of relying on external GPS infrastructure, the system creates a local copy of the navigation reference frame by detecting and tracking environmental features (landmarks, structures) in the scene. This feature-based local coordinate system serves as a substitute for GPS, enabling high-precision relative positioning that achieves within 2 meters accuracy for inspection applications.
3Adaptability or versatility
If the AV operates in GPS-denied environments, then it can perform civilian inspection tasks, but navigation accuracy deteriorates without external reference systems
Solution Approach 1:
The navigation system becomes self-sufficient in GPS-denied environments by using its own onboard cameras to detect and track environmental features. The perceptive navigation subsystem extracts positional information directly from the environment without requiring external infrastructure, allowing the AV to autonomously navigate and maintain accuracy in diverse settings including underground, indoor, and bridge environments.
Data Source
AI summary
A navigation system and method for an autonomous vehicle (AV) that includes a system controller and an environmental sensor. The system controller determines a fixed local frame (LCF) having a coordinate system originating at a fixed location in three-dimensional space. Additionally, the system controller determines an AV positional pose in LCF coordinates, which identifies an AV position in three-dimensional space and an AV orientation in three-dimensional space. An environmental sensor detects an asset feature of an asset moving within three-dimensional space. The system controller then identifies an asset feature frame (AFF) having a coordinate system originating at the asset feature. A localization module determines the AV positional pose in the AFF coordinates. The system controller then dynamically transforms the AV position pose from the LCF coordinates to the AFF coordinates and generates a motion control command based on the AV positional pose in the AFF coordinates.