Forward Camera Pose Determination via Driver Gaze Mapping

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In vehicle monitoring systems, accurately mapping the driver's gaze from a driver-facing camera to a forward-facing camera is challenging due to vastly different camera poses and non-overlapping fields of view.

Innovation Solution

A method that captures images of the driver's face and the forward road scene from respective cameras, processes the driver's gaze direction data, statistically collates it into a frequency distribution, identifies peaks associated with reference points in the forward road scene, and correlates these points to determine the camera pose of the forward-facing camera.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If multiple cameras are positioned at different positions and orientations to monitor a scene from different perspectives, then the ability to track and map object positions between camera views is improved, but the device complexity increases

Engineering Contradiction:
Improveability to track and map object positionsVSAvoidcamera positioning complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary coordinate transformation system that maps between driver-facing camera coordinates and forward-facing camera coordinates through a vehicle-centric reference frame. This mediator enables position tracking across non-overlapping camera views without requiring complex direct camera-to-camera calibration, thus resolving the contradiction between mapping capability and system complexity.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Ease of operation

If driver-facing camera and forward-facing camera have vastly different camera poses, then the driver's gaze can be captured relative to the driver's view, but the ability to project gaze onto the forward road scene is degraded

Engineering Contradiction:
Improvegaze capture easeVSAvoidgaze projection accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The patent resolves the gaze projection problem by introducing an additional dimensional layer - a vehicle-centric coordinate system that serves as an intermediate reference frame. Instead of directly mapping between two differently oriented camera views, the solution transforms gaze data from the driver-facing camera through the vehicle coordinate system to the forward-facing camera, effectively adding a dimensional bridge that preserves measurement precision while maintaining ease of gaze capture.

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

3Area of stationary object

If cameras are positioned to capture different fields of view (driver view and road view), then the monitoring coverage is improved, but the overlapping or adjacent fields of view necessary for accurate mapping are lost

Engineering Contradiction:
Improvemonitoring coverage areaVSAvoidgaze mapping accuracy
Core Design Contradiction:
Area of stationary objectVSMeasurement precision

Solution Approach 1:

The patent employs a vehicle-centric coordinate system as an intermediary reference frame that decouples the mapping accuracy from the requirement of overlapping camera fields of view. By transforming all measurements into this common vehicle-based coordinate system, the system maintains high mapping accuracy across non-overlapping views, thus resolving the contradiction between maximizing monitoring coverage and maintaining mapping precision.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3479356B1System and method for identifying a camera pose of a forward facing camera in a vehicle
Publication Date: 2025.04.16 SEEING MACHINES
  • EP3479356B1 patent drawingFigure 1
  • EP3479356B1 patent drawingFigure 2
  • EP3479356B1 patent drawingFigure 3

AI summary

A computerized method of determining a camera pose of a forward facing camera in a vehicle scene, including: capturing images of a vehicle driver's face from a driver facing camera and images of the forward road scene from a forward facing camera; processing the images of the driver's face to derive gaze direction data in a vehicle frame of reference; the gaze direction data are statistically collated into a frequency distribution of gaze angles; peaks in the frequency distribution are identified and associated with reference points in the images of the forward road scene to determine one or more reference gaze positions in the vehicle reference frame; the one or more reference gaze positions are correlated with a position of the reference points in the forward facing camera reference frame to determine a camera pose of a forward facing camera in the vehicle frame of reference.