Vehicle Travel Control Using Split State Matrices

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In complex traffic scenarios, existing vehicle control methods face high calculation complexity and dimensionality issues due to large matrices representing vehicle states, leading to inefficiencies in determining control policies, especially in multi-vehicle interaction scenarios.

Innovation Solution

The method involves obtaining a target matrix based on vehicle and environmental information, splitting it into sub-matrices using nonparametric estimation, and performing iteration calculations to obtain updated matrix elements, which are then used to determine target travel control information considering surrounding vehicle interactions, thereby reducing calculation complexity and ensuring safety.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a large target matrix is used to represent vehicle states in complex traffic scenarios, then the control policy can capture more comprehensive state information, but the calculation complexity and dimensionality increase significantly

Engineering Contradiction:
Improvestate representation accuracyVSAvoidcalculation complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the large target matrix into multiple sub-matrices based on different state dimensions (position, velocity, acceleration). Each sub-matrix handles a specific aspect of vehicle state, reducing the computational burden of processing the entire state space while maintaining comprehensive state representation capability

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transforms the high-dimensional matrix operation problem into a lower-dimensional problem by using nonparametric estimation methods. Instead of directly computing with the full target matrix, the system estimates transition probabilities through sampling and kernel density estimation, effectively reducing dimensionality while preserving state transition information

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

2Measurement precision

If iteration calculations are performed to update matrix elements for accurate probability estimation, then the control policy accuracy improves, but the computation time increases

Engineering Contradiction:
Improveprobability estimation accuracyVSAvoidcomputation time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent implements iterative updates of matrix elements at discrete time steps rather than continuously. The iteration process updates transition probabilities periodically based on new observations, balancing accuracy improvement with computational efficiency by avoiding constant recalculation

Inventive Principle:
Principle #19Periodic action

Solution Approach 2:

The patent pre-computes certain matrix elements and transition probabilities that are less likely to change, storing them for reuse. This preliminary computation reduces the need for repeated calculations during real-time control, decreasing computation time while maintaining accuracy for dynamically changing state transitions

Inventive Principle:
Principle #10Preliminary action

3Reliability

If the target matrix includes interactions with surrounding vehicles in multi-vehicle scenarios, then the safety of the target vehicle is improved, but the matrix dimensionality and calculation complexity increase

Engineering Contradiction:
Improvevehicle safetyVSAvoidmatrix dimensionality
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments the interaction model by creating separate sub-matrices for different surrounding vehicles and interaction types (collision risk, lane changing, merging). This segmentation allows the system to process vehicle interactions independently and combine results, reducing the overall matrix dimensionality while capturing multi-vehicle interaction effects for improved safety

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts only the critical interaction parameters from the full vehicle state space when modeling multi-vehicle interactions. Instead of including all state variables for each surrounding vehicle, the system extracts relevant features (relative position, relative velocity, interaction type) to build compact interaction sub-matrices that maintain safety analysis capability with reduced dimensionality

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP3919337B1Control vehicle
Publication Date: 2023.11.29 BEIJING SANKUAI ONLINE TECH CO LTD
  • EP3919337B1 patent drawingFigure 1~2
  • EP3919337B1 patent drawingFigure 3~4
  • EP3919337B1 patent drawingFigure 5

AI summary

Vehicle control is provided, : obtaining vehicle information of a target vehicle and environmental information of a reference environment in which the target vehicle is located; obtaining a target matrix based on the vehicle information and the environmental information; splitting the target matrix to obtain a plurality of sub-matrices; and obtaining target travel control information of the target vehicle based on matrix elements in the sub-matrices and travel control information of a surrounding vehicle of the target vehicle.