Camera Motion Scale Estimation via Ground Calibration Plane

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing sensor systems for calculating camera motion in vehicles face challenges in estimating scale without features on the ground, particularly in nonplanar ground scenarios, leading to error propagation and instability in motion estimation.

Innovation Solution

A sensor device that estimates camera motion scale by analyzing the displacement of points and projected lines in a ground calibration plane, allowing for dynamic motion estimation and object tracking without requiring ground features, and isolating rotational and translational components of motion.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If ground calibration plane is used for motion estimation, then motion scale can be calculated, but the ground plane changes over time and may no longer be flat causing projection errors

Engineering Contradiction:
Improvemotion scale calculationVSAvoidground calibration plane stability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The system dynamically adapts to changing ground conditions by continuously re-estimating the ground calibration plane parameters. Instead of relying on a static pre-calibrated plane, the system updates the plane parameters based on current image data, allowing it to handle non-planar and dynamically changing ground surfaces while maintaining accurate motion scale calculation

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system performs self-calibration by automatically estimating and updating the ground calibration plane parameters using features detected in the current image sequence. This self-service mechanism eliminates the need for external calibration tools or manual intervention, allowing the system to adapt to changing ground conditions autonomously

Inventive Principle:
Principle #25Self-service

2Measurement precision

If features on the ground are required for scale estimation, then motion can be calculated, but the system fails in off-road scenarios without ground features

Engineering Contradiction:
Improvemotion scaleVSAvoidground condition adaptability
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The system achieves multi-functionality by being able to operate in both planar ground scenarios (like roads) and non-planar off-road scenarios. The ground calibration plane estimation method is designed to work with any detectable features regardless of ground type, making the system universal across different terrains and conditions without requiring separate processing paths

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

3Measurement precision

If multiple sensors are fused for motion estimation, then global motion can be obtained, but errors are propagated through the feedback loop

Engineering Contradiction:
Improveglobal motion estimationVSAvoiderror propagation
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The system extracts and uses only the essential information needed for motion estimation - specifically the ground calibration plane parameters and feature displacements - while discarding redundant or potentially error-prone data from other sensors. This selective extraction approach obtains global motion estimation without incorporating error-prone feedback loops from multiple sensor fusions

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP3789961B1Sensor device for calculating a motion of a camera in a vehicle
Publication Date: 2023.06.14 CONTINENTAL AUTONOMOUS MOBILITY GERMANY GMBH
  • EP3789961B1 patent drawingFigure 1~2
  • EP3789961B1 patent drawingFigure 3~4
  • EP3789961B1 patent drawingFigure 5~6

AI summary

The invention is directed to a sensor device (100) for calculating a motion of a camera (102) in a vehicle (200), which is capable of estimating the motion of the camera (102) with scale, which can be beneficial in off-road scenarios.