Autonomous Vehicle Behavior Prediction Error Prioritization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Autonomous vehicles face challenges in accurately predicting the behavior of objects in their environment, leading to suboptimal driving decisions due to incorrect object behavior predictions, which can impact safety and efficiency.

Innovation Solution

A system that compares predicted behavior data with observed behavior data to identify incorrect object behavior predictions, prioritizing them based on relevancy by assessing spatial and temporal overlaps, and initiates operations to address these errors, ensuring more critical errors are corrected first.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the autonomous vehicle uses behavior prediction systems to identify object behaviors, then the driving decisions can be made based on predicted behaviors, but incorrect predictions lead to suboptimal decisions impacting safety and efficiency

Engineering Contradiction:

Improveprediction accuracyVSAvoidsafety risks from incorrect predictions

Core Design Contradiction:

ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The system compares predicted behavior data with observed behavior data to identify incorrect predictions, creating a feedback loop that continuously improves prediction accuracy by learning from discrepancies between predicted and actual object behaviors

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system proactively identifies and addresses incorrect predictions before they lead to harmful outcomes by continuously monitoring prediction accuracy and initiating corrections in advance

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If the system addresses all incorrect behavior predictions, then prediction accuracy improves, but processing time and computational resources increase

Engineering Contradiction:

Improvebehavior prediction accuracyVSAvoidtime to process and correct errors

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system applies different processing priorities to different incorrect predictions based on their relevancy indicators, focusing computational resources on high-priority errors that most impact safety and efficiency while using fewer resources for low-priority errors

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The system continuously monitors and re-evaluates prediction errors, adjusting processing priorities based on changing conditions and accumulating evidence about which types of errors are most critical

Inventive Principle:

Principle #23Feedback

3Reliability

If the system prioritizes correction of critical errors based on spatial and temporal overlap analysis, then safety is improved, but system complexity increases

Engineering Contradiction:

Improvesafety of navigationVSAvoidcomplexity of error prioritization system

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system uses spatial and temporal overlap parameters to quantify the criticality of different prediction errors, transforming a complex qualitative assessment into a more manageable quantitative analysis based on measurable parameters

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS11904886B1Modifying autonomous vehicle behavior prediction based on behavior prediction errors

Publication Date: 2024.02.20 WAYMO LLC

AI summary

A system includes a memory device and a processing device, operatively coupled to the memory device, to obtain a set of predicted behavior data including data indicative of one or more predicted object behaviors for one or more respective objects in an environment of an autonomous vehicle, obtain a set of observed behavior data including data indicative of one or more observed object behaviors of the one or more objects, determine, based on a comparison of the set of predicted behavior data and the set of observed behavior data, whether a set of incorrect object behavior predictions exists within the set of predicted behavior data, and upon determining that the set of incorrect object behavior predictions exists, initiate, based on at least a generated subset of the set of incorrect object behavior predictions, one or more operations to address the set of incorrect object behavior predictions.