Autonomous Vehicle Data Capture Scheduling for Limited Onboard Storage

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Autonomous vehicles face challenges in efficiently collecting and storing data due to limited storage capacity, lacking granular instructions for when and where to collect image data, and being unable to optimize data collection based on operating conditions, leading to inadequate training data for improved autonomy algorithms.

Innovation Solution

A data-driven system that identifies conditions in roadways for optimized data collection by generating condition capture instructions for autonomous vehicles to collect data at specific times and locations, reducing unnecessary data capture, and communicating these instructions to improve storage efficiency and extend travel time.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If autonomous vehicles continuously collect and store all onboard data including raw images and high quality images, then the training data for autonomy algorithms is comprehensive, but the storage capacity is quickly exhausted and bandwidth is consumed

Engineering Contradiction:

Improvequantity of training dataVSAvoidstorage capacity

Core Design Contradiction:

Quantity of substanceVSVolume of stationary object

Solution Approach 1:

The system performs preliminary analysis of roadway conditions using prediction models before data collection. It identifies specific conditions (e.g., adverse weather, complex intersections, pedestrian activity) that warrant data collection in advance, allowing the AV to selectively capture only relevant data rather than continuously storing all data

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system applies different data collection strategies to different roadway conditions. Instead of uniform continuous collection, it adjusts collection intensity and types based on local condition characteristics - collecting comprehensive data only when prediction models identify high-value conditions, and reducing or eliminating collection during normal conditions

Inventive Principle:

Principle #3Local quality

2Adaptability or versatility

If autonomous vehicles collect data at all times and locations, then the training data covers all possible scenarios, but the storage is filled quickly and travel time is reduced

Engineering Contradiction:

Improvecoverage of training scenariosVSAvoidtravel time

Core Design Contradiction:

Adaptability or versatilityVSDuration of action of moving object

Solution Approach 1:

The system uses prediction models to pre-identify high-value data collection opportunities along the AV's route. By analyzing roadway conditions, traffic patterns, and environmental factors in advance, the system creates a optimized collection schedule that ensures comprehensive scenario coverage while minimizing unnecessary data collection during low-value periods

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

Instead of continuous data collection, the system implements periodic collection triggered by predicted high-value conditions. The prediction models continuously monitor roadway conditions and trigger data collection only when specific adverse or complex conditions are forecasted, creating an on-demand periodic collection pattern rather than continuous operation

Inventive Principle:

Principle #19Periodic action

3Ease of operation

If autonomous vehicles lack granular instructions for data collection, then the system is simple to operate, but the data collection is not optimized and storage efficiency is poor

Engineering Contradiction:

Improvesimplicity of data collection operationVSAvoidstorage efficiency

Core Design Contradiction:

Ease of operationVSLoss of energy

Solution Approach 1:

The system implements self-service through automated prediction models that independently analyze roadway conditions and generate data collection instructions without human intervention. The models automatically identify when and where to collect data based on pre-defined criteria for high-value conditions, eliminating the need for manual operational complexity while maximizing storage efficiency

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system incorporates feedback loops where collected data is continuously analyzed to refine the prediction models. The models learn from actual data collection outcomes and adjust their predictions to better identify high-value conditions, creating a self-improving system that becomes increasingly efficient at optimizing data collection without additional operational complexity

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS12351197B2System, method, and computer program product for data-driven optimization of onboard data collection

Publication Date: 2025.07.08 FORD GLOBAL TECH LLC

AI summary

Provided are systems, methods, and computer program products for data-driven optimization of onboard data collection, comprising identifying a condition in a roadway associated with an operation of autonomous vehicles (AVs) that may be further optimized to improve performance of one or more AVs; and generating condition capture instructions to communicate to the AVs, wherein the condition capture instructions comprise one or more parameters of a storage request in one or more roadways predicted to exhibit the condition in the roadway; and when at least one instruction in the one or more condition capture instructions is received by an AV, the at least one instruction is configured to cause the AV to collect condition information in the roadway at a time when the condition is predicted to be present in the roadway.