Neural Network Vehicle Trajectory Control via Time-Series Signals

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current vehicle control systems for autonomous or semi-autonomous vehicles require extensive computational resources and time for object detection and dynamic map creation, making them inefficient for real-time motion control, especially when dealing with diverse environments and scenarios.

Innovation Solution

A neural network system that transforms time-series environmental signals into reference trajectories, decoupling the process from specific vehicle dynamics and allowing offline training for online reuse, reducing the need for object detection and dynamic maps by focusing on relevant information for motion control.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If object detection and dynamic map creation are performed using traditional methods, then the vehicle can understand the environment and make safe decisions, but the process consumes excessive computational resources and time

Engineering Contradiction:
Improveenvironment understanding accuracyVSAvoidreal-time control efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The neural network is trained offline in advance on diverse driving scenarios and environments, storing learned patterns and representations in its weights and parameters. This preliminary training enables the network to process new sensor inputs efficiently during real-time operation without requiring extensive computational resources for object detection and map creation

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent replaces traditional mechanical vision systems (object detection algorithms, dynamic map building processes) with a neural network-based system. Instead of explicitly detecting objects and constructing maps, the neural network directly processes sensor inputs to generate control commands, substituting complex computational mechanics with learned behavioral patterns

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If comprehensive object detection and recognition are performed, then the vehicle can accurately understand the environment, but the computational complexity and processing time increase significantly

Engineering Contradiction:
Improveobject detection accuracyVSAvoidcomputational resource requirements
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts and removes the intermediate steps of object detection, recognition, and dynamic map creation from the control pipeline. The neural network is trained to bypass these complex processing stages and directly map sensor inputs to control commands, extracting only the essential information needed for safe vehicle operation

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The neural network is designed as a universal system that handles multiple functions simultaneously: it processes various sensor inputs (camera, LIDAR, radar), understands diverse environments (different countries, road types, weather conditions), and generates appropriate control commands. This multi-functional approach eliminates the need for separate specialized modules for each detection and control task

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

Data Source

PatentEP3535636B1Method and system for controlling vehicle
Publication Date: 2020.09.23 MITSUBISHI ELECTRIC CORP
  • EP3535636B1 patent drawingFigure 1A
  • EP3535636B1 patent drawingFigure 1B
  • EP3535636B1 patent drawingFigure 1C

AI summary

A method and a system generate a time-series signal indicative of a variation of the environment in vicinity of the vehicle with respect to a motion of the vehicle and submit the time-series signal to the neural network to produce a reference trajectory as a function of time that satisfies time and spatial constraints on a position of the vehicle. The neural network is trained in to transform time-series signals to reference trajectories of the vehicle. The motion trajectory tracking the reference trajectory while satisfying constraints on the motion of the vehicle is determined and the motion of the vehicle is controlled to follow the motion trajectory.