Vehicle Motion Detection Using Optical Flow for Adaptive ZUPT

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing zero velocity update (ZUPT) techniques for vehicles struggle with unreliable motion detection, particularly in low-cost inertial sensors, noisy measurements, and inflexible threshold comparisons, leading to inaccurate navigation system corrections.

Innovation Solution

A machine learning-based system using optical flow algorithms and artificial neural networks to analyze image sequences, combined with depth estimation, for accurate and adaptive motion detection of vehicles, eliminating the need for manual threshold settings.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If manually tuned thresholds are used to determine vehicle motion, then the system can operate with simple logic, but the system becomes inflexible and inaccurate requiring frequent calibration

Engineering Contradiction:
Improveoperation simplicityVSAvoidenvironmental adaptability
Core Design Contradiction:
Ease of operationVSAdaptability or versatility

Solution Approach 1:

The patent implements dynamic threshold adjustment by training a neural network to adapt thresholds based on environmental conditions and sensor characteristics. The system transitions from static manually-tuned thresholds to dynamic thresholds that automatically adjust to varying operating conditions, resolving the contradiction between operational simplicity and environmental adaptability

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes the parameter of motion detection from fixed threshold values to adaptive threshold values determined by machine learning models. This parameter transformation allows the system to maintain simplicity while gaining adaptability through data-driven threshold selection that responds to changing environmental conditions

Inventive Principle:
Principle #35Parameter changes

2Ease of manufacture

If low-cost inertial sensors are used for motion detection, then the system cost is reduced, but the measurements become too noisy to reliably trigger ZUPT

Engineering Contradiction:
Improvesystem costVSAvoidmotion detection reliability
Core Design Contradiction:
Ease of manufactureVSReliability

Solution Approach 1:

The patent introduces an intermediary neural network processing layer between the low-cost inertial sensors and the ZUPT trigger decision. This intermediary processes and filters the noisy sensor data, extracting meaningful motion information while suppressing noise, thereby enabling reliable ZUPT triggering with low-cost sensors

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces traditional mechanical threshold-based motion detection logic with a machine learning-based detection system. This substitution transforms the approach from direct sensor threshold comparison to intelligent pattern recognition, enabling reliable motion detection despite sensor noise while maintaining low system cost

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

3Ease of operation

If imaging techniques with feature extraction are used for ZUPT, then motion detection can be achieved, but accuracy decreases depending on the application

Engineering Contradiction:
Improvedetection capabilityVSAvoidmotion detection accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The patent develops a universal motion detection framework using neural networks that can process multiple types of input data (inertial sensor data, image data, or combinations) and adapt to different applications. This multi-functional approach maintains ease of operation while improving measurement precision through robust feature learning that works across diverse scenarios

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

4Device complexity

If traditional threshold comparison methods are used, then the system remains simple, but frequent calibration is required maintaining complexity

Engineering Contradiction:
Improvesystem complexityVSAvoidcalibration time
Core Design Contradiction:
Device complexityVSLoss of time

Solution Approach 1:

The patent implements self-service through automated neural network-based threshold adjustment that eliminates the need for manual calibration. The system automatically adapts to changing conditions and optimizes its parameters without human intervention, reducing both complexity and calibration time while maintaining detection accuracy

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP3955209B1Machine learning and vision-based approach to zero velocity update object detection
Publication Date: 2026.03.18 HONEYWELL INTERNATIONAL INC
  • EP3955209B1 patent drawingFigure 1
  • EP3955209B1 patent drawingFigure 2
  • EP3955209B1 patent drawingFigure 3

AI summary

Techniques for detecting motion of a vehicle are disclosed. Optical flow techniques are applied to the entirety of the received images from an optical sensor mounted to a vehicle. Motion detection techniques are then imposed on the optical flow output to remove image portions that correspond to objects moving independent from the vehicle and determine the extent, if any, of movement by the vehicle from the remaining image portions. Motion detection can be performed via a machine learning classifier. In some aspects, motion can be detected by extracting the depth of received images in addition to optical flow. In additional or alternative aspects, the optical flow and/or motion detection techniques can be implemented by at least one artificial neural network.