Vehicle Sensor Data Coding for High-Throughput Wireline Links

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Autonomous vehicles face challenges in processing high volumes of sensor data in real-time due to high data throughput requirements, leading to outdated data processing and increased material costs.

Innovation Solution

A device that generates coded packets from multiple data streams and transmits them through a wireline network to AI accelerator modules and memory units, utilizing vector packet coding to increase data throughput without increasing material costs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If sensors generate high amount of data and stream to processing elements, then data processing capability is improved, but time expenditure increases and material costs increase

Engineering Contradiction:
Improvedata processing capabilityVSAvoidtime expenditure
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent segments data streams by type (e.g., camera data, LIDAR data, radar data) and routes them through different processing paths. Each data type is handled by specialized processing elements optimized for that specific data format, reducing overall processing time and enabling parallel processing of multiple data streams simultaneously.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a new dimension of data handling by implementing a multi-layer processing architecture with different processing priorities. Time-critical data (e.g., obstacle detection) is processed in a high-priority path with lower latency, while less time-sensitive data (e.g., environmental mapping) is processed in a lower-priority path, effectively adding a temporal dimension to data processing.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Productivity

If sensors generate high amount of data and stream to processing elements, then data processing capability is improved, but material costs increase

Engineering Contradiction:
Improvedata processing capabilityVSAvoidmaterial costs
Core Design Contradiction:
ProductivityVSQuantity of substance

Solution Approach 1:

The patent implements universal processing elements that can handle multiple data types through software configuration rather than requiring dedicated hardware for each sensor type. A single processing element can be dynamically allocated to process camera data, LIDAR data, or radar data based on real-time needs, reducing the total number of processing elements required and thereby reducing material costs.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent merges multiple data streams of the same type into consolidated processing pipelines. Instead of having separate processing paths for each individual sensor output, multiple camera feeds are combined into a single processing stream, and multiple LIDAR streams are aggregated, reducing redundancy in processing infrastructure and lowering material costs.

Inventive Principle:
Principle #5Merging (Combining)

3Loss of information

If data processing is performed in real-time, then data currency is improved, but processing speed requirements increase leading to higher costs

Engineering Contradiction:
Improvedata currencyVSAvoidprocessing speed
Core Design Contradiction:
Loss of informationVSProductivity

Solution Approach 1:

The patent performs preliminary processing actions on data streams before they reach the main processing elements. Data pre-processing steps such as filtering, noise reduction, and feature extraction are executed in advance, reducing the computational burden on real-time processing elements and enabling faster processing without sacrificing data currency.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4015335B1Device for a vehicle
Publication Date: 2024.10.09 INTEL CORP
  • EP4015335B1 patent drawingFigure 1
  • EP4015335B1 patent drawingFigure 2
  • EP4015335B1 patent drawingFigure 3A

AI summary

A device for a vehicle may include a first wireline interface configured to receive a first data stream from a first sensor having a first sensor type for perceiving a surrounding of the vehicle, the first data stream including raw sensor data detected by the first sensor; a second wireline interface configured to receive a second data stream from a second sensor having a second sensor type for perceiving the surrounding of the vehicle, the second data stream including raw sensor data detected by the second sensor; one or more processors configured to generate a coded packet including the received first data stream and the received second data stream by employing vector packet coding on the first data stream and the second data stream; and an output wireline interface configured to transmit the generated coded packet to one or more target units of the vehicle.