Reference Graph Scene Context for Faster Autonomous Vehicle Decisions

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Generating scene context representative of dependencies between all agents and environment features in autonomous vehicles is computationally expensive and time-consuming.

Innovation Solution

Pre-computing scene context data using a reference graph that includes pre-generated feature vectors for static objects, allowing for faster scene context generation by accessing pre-computed data during vehicle operations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If scene context is generated from sensor data representing all agents and environment features, then completeness of scene context is improved, but computational cost and time consumption increase

Engineering Contradiction:

Improvecompleteness of scene contextVSAvoidtime consumption for scene context generation

Core Design Contradiction:

Loss of informationVSLoss of time

Solution Approach 1:

The system pre-generates feature vectors for static environment features before autonomous vehicle operations. These feature vectors are stored in a reference graph structure, allowing the vehicle to quickly access pre-computed scene context data without performing full computational analysis during operation, thus reducing time consumption while maintaining completeness

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The scene context generation is divided into two segments: static environment features are processed offline to create feature vectors, while dynamic agent features are processed online. This segmentation allows computationally intensive static feature processing to be done in advance, separating it from time-critical operational decision-making

Inventive Principle:

Principle #1Segmentation

2Reliability

If scene context represents dependencies between all agents and environment features, then accuracy of operational decisions is improved, but computational resources required increase

Engineering Contradiction:

Improveaccuracy of operational decisionsVSAvoidcomputational resources

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

Feature vectors for static environment features are pre-computed and stored in the reference graph before the autonomous vehicle needs to make operational decisions. This preliminary computation captures spatial relationships and dependencies in advance, reducing the computational burden during actual operation while preserving accuracy

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system creates simplified representations (feature vectors) that copy essential characteristics of complex environment features. These feature vectors preserve the necessary information for accurate decision-making while requiring significantly fewer computational resources to process than the original detailed sensor data

Inventive Principle:

Principle #26Copying

Data Source

PatentUS20250206344A1System for Generating Scene Context Data Using a Reference Graph

Publication Date: 2025.06.26 ZOOX INC

AI summary

Techniques for improving operational decisions of an autonomous vehicle are discussed herein. In some cases, a system may generate reference graphs associated with a route of the autonomous vehicle. Such reference graphs can comprise precomputed feature vectors based on grid regions and/or lane segments. The feature vectors are usable to determine scene context data associated with static objects to reduce computational expenses and compute time.