Autonomous Vehicle Control Using State-Differential Rules
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous vehicle systems face challenges in training a single end-to-end model for making driving decisions, as they require processing various environmental data inputs from cameras and sensors to determine control operations effectively.
Innovation Solution
An operational model processes camera data to determine the current and predicted environmental states, calculates a differential with a previously predicted state, and uses a rules module to determine control operations for the autonomous vehicle, incorporating redundant power and data fabrics for reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single end-to-end model is used for making driving decisions, then the system complexity is reduced, but the training feasibility and reliability deteriorate due to the inability to effectively process various environmental data inputs from cameras and sensors
Solution Approach 1:
The patent segments the autonomous driving decision-making system into multiple specialized modules: an operational model for processing camera data and determining environmental states, a rules module for processing sensor data and applying safety rules, and a motion model for generating control operations. This segmentation allows each module to be trained independently on specific data types and tasks, improving training feasibility while maintaining manageable system complexity.
Solution Approach 2:
The patent introduces intermediary components including a motion estimation module that generates motion vectors from camera data, and a normalization module that standardizes input data. These intermediaries bridge the gap between different data sources (cameras and sensors) and the decision-making modules, enabling effective processing of diverse environmental data inputs without requiring a single complex end-to-end model.
2Reliability
If multiple data sources (cameras and sensors) are processed to determine control operations, then the reliability and accuracy of driving decisions are improved, but the device complexity and processing requirements increase
Solution Approach 1:
The system segments processing of different data sources into dedicated modules: the operational model handles camera data for environmental state determination, while the rules module processes sensor data for safety rule application. This segmentation allows each module to optimize for its specific data type without increasing overall system complexity.
Solution Approach 2:
The motion estimation module serves multiple functions by generating motion vectors that are used both by the operational model for predicting environmental states and by the rules module for safety assessments. This multi-functionality reduces redundancy and maintains reliability while controlling processing complexity.
Data Source
AI summary
Determining control operations for an autonomous vehicle may include receiving, by an operational model, an operational model input based on camera data from one or more cameras of an automated vehicle; determining, by the operational model, based on the input, a current environmental state and a predicted environmental state; providing, to a rules module, a differential between the current environmental state and a previously predicted environmental state; and determining, based on the rules module, one or more control operations for the automated vehicle.