Vehicle Driving Control Using Multi-Chip Index Aggregation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing autonomous driving technologies struggle to achieve high-accuracy, real-time driving control using vast amounts of sensor data due to limitations in computational power and data processing efficiency, particularly at higher levels of autonomy beyond Level 5.

Innovation Solution

An information processing device employing multiple chips and an anchor chip to perform multivariate analysis via deep learning, aggregating index values from various sensors at ultra-high speeds (billionths of a second) to enable Level 6 autonomous driving, with dynamic load balancing and priority-based weighting to optimize driving control.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If multiple chips are used to increase computational power for processing vast sensor data, then driving control accuracy is improved, but device complexity increases

Engineering Contradiction:
Improvedriving control accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system divides the computational task into multiple independent chips, each processing specific sensor data streams. This segmentation allows parallel processing of vast amounts of sensor data while maintaining manageable complexity through modular architecture, where each chip handles a discrete portion of the overall computation required for Level 6 autonomous driving control

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Multiple chips are merged into a unified processing system with a central control unit that aggregates results from all chips. This merging enables the system to achieve the computational power necessary for high-accuracy driving control while presenting a cohesive interface to the vehicle's driving systems, effectively managing complexity through integrated architecture

Inventive Principle:
Principle #5Merging (Combining)

2Productivity

If multivariate analysis by deep learning is performed using multiple chips, then productivity is improved, but use of energy increases

Engineering Contradiction:
Improvedata processing speedVSAvoidenergy consumption
Core Design Contradiction:
ProductivityVSUse of energy by moving object

Solution Approach 1:

The deep learning computation is segmented across multiple chips, allowing parallel processing of sensor data through multivariate analysis. This segmentation increases productivity by enabling simultaneous processing of multiple data streams while distributing energy consumption across the chip set, preventing any single chip from becoming an energy bottleneck

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system performs periodic aggregation of results from multiple chips at predetermined intervals, allowing each chip to process data in discrete cycles rather than continuous operation. This periodic action optimizes energy efficiency by enabling power management during aggregation intervals while maintaining high overall processing throughput through parallel periodic computations

Inventive Principle:
Principle #19Periodic action

3Speed

If index values are aggregated at ultra-high speeds in units of billionth of a second, then driving control responsiveness is improved, but loss of energy increases

Engineering Contradiction:
Improveprocessing speedVSAvoidenergy loss
Core Design Contradiction:
SpeedVSLoss of energy

Solution Approach 1:

The anchor chip aggregates index values from multiple chips at predetermined intervals in units of billionth of a second, creating a periodic aggregation rhythm. This approach achieves ultra-high speed responsiveness by performing aggregations only at necessary intervals rather than continuously, significantly reducing energy loss compared to continuous real-time processing while maintaining the speed required for Level 6 autonomous driving control

Inventive Principle:
Principle #19Periodic action

Solution Approach 2:

Multiple chips perform preliminary processing and generate index values in parallel before the anchor chip performs the final aggregation. This preliminary action allows the system to prepare data for ultra-high speed aggregation without requiring all chips to operate at full capacity simultaneously, reducing overall energy loss while achieving the required processing speed through coordinated parallel-preparation and periodic-aggregation cycles

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4620767A1Information processing device, program, and information processing system
Publication Date: 2025.09.24 SOFTBANK GROUP CORP
  • EP4620767A1 patent drawingFigure 1
  • EP4620767A1 patent drawingFigure 2
  • EP4620767A1 patent drawingFigure 3

AI summary

An information processing device disclosed herein includes: an information acquisition unit capable of acquiring plural pieces of information related to a vehicle; plural chips that are provided for each of the plural pieces of information and infer an index value from predetermined information among the plural pieces of information; and a driving control unit that executes driving control of the vehicle on a basis of the plural index values.