Environment Map Fusion for Robust Object Pose Tracking

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing augmented reality systems face challenges in robustly tracking objects, especially when they become small, partially occluded, or move out of the camera's field of view, leading to inaccurate pose estimation and tracking jitter.

Innovation Solution

The system utilizes a dynamically generated 3D environment map data to estimate the target pose, fusing it with the environment pose to improve tracking robustness and extend the trackable range beyond the visible limits, even for moving targets.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional target-only tracking methods are used, then the system is simple to implement, but tracking reliability deteriorates when the target is small, occluded, or out of view

Engineering Contradiction:
Improvetracking reliabilityVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent combines target-based tracking with environment-based SLAM tracking into a unified system. The environment map is built by integrating target map data with dynamically captured environment features, creating a composite tracking system that leverages both target appearance and surrounding environment geometry to maintain tracking reliability under various conditions.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The environment map serves as an intermediary structure that connects the camera to the target. When direct target tracking fails, the system uses the pre-built environment map to estimate camera pose and infer target pose, acting as a mediator that maintains tracking continuity even when the target is occluded or out of view.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If target-based tracking is used, then the system has simple data requirements, but measurement precision deteriorates for distant or small targets

Engineering Contradiction:
Improvepose estimation precisionVSAvoidtracking range
Core Design Contradiction:
Measurement precisionVSLength of stationary object

Solution Approach 1:

The patent transitions from 2D image-space target tracking to 3D environment-space tracking. By building a three-dimensional environment map from multiple views and using it for pose estimation, the system extends accurate tracking to distant targets that would be too small or feature-poor in the camera view, leveraging the additional spatial dimension of the environment structure.

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

3Adaptability or versatility

If pre-built SLAM maps are used, then the system has fast runtime performance, but adaptability deteriorates for dynamic environments or moving targets

Engineering Contradiction:
Improveenvironment adaptabilityVSAvoidruntime performance
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The system performs preliminary environment mapping during an initialization phase, capturing and processing environment features before runtime tracking begins. This pre-built environment map enables fast runtime performance, while the system maintains adaptability by using the flexible SLAM framework that can accommodate dynamic changes and moving targets during operation.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3053099B1Object tracking based on dynamically built environment map data
Publication Date: 2026.02.25 QUALCOMM INC
  • EP3053099B1 patent drawingFigure 1A~1B
  • EP3053099B1 patent drawingFigure 2
  • EP3053099B1 patent drawingFigure 3

AI summary

A computer-implemented method of tracking a target object in an object recognition system includes acquiring a plurality of images with a camera. The method further includes simultaneously tracking the target object and dynamically building environment map data from the plurality of images. The tracking of the target object includes attempting to estimate a target pose of the target object with respect to the camera based on at least one of the plurality of images and based on target map data. Next, the method determines whether the tracking of the target object with respect to the camera is successful. If not, then the method includes inferring the target pose with respect to the camera based on the dynamically built environment map data. In one aspect the method includes fusing the inferred target pose with the actual target pose even if tracking is successful to improve robustness.