Host Pose Correction Using Reference Object Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In robot localization, pose drifting errors accumulate over time, leading to inaccuracies in tracking the pose of robots, especially in long-term tracking scenarios.

Innovation Solution

A method for pose correction using a host device that obtains images, determines the presence of reference objects, calculates their relative positions, and corrects the host's pose based on predetermined reference poses stored in look-up tables.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Duration of action of stationary object

If pose tracking is performed for a long time, then continuous pose estimation is achieved, but pose drifting errors accumulate leading to reduced accuracy

Engineering Contradiction:

Improvetracking durationVSAvoidpose accuracy

Core Design Contradiction:

Duration of action of stationary objectVSMeasurement precision

Solution Approach 1:

The patent implements a feedback mechanism by detecting reference objects in the environment and using their known positions to calculate correction values for the accumulated pose drift. The system continuously monitors pose estimation accuracy by comparing estimated positions with actual reference object positions, and applies corrections based on the detected deviations, thereby maintaining long-term tracking accuracy.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent changes the operational parameters of the tracking system by switching between continuous pose estimation mode and correction mode. When a reference object is detected, the system transitions to correction mode where it calculates correction values based on the reference object's known position, then applies these corrections to reset the accumulated drift errors, effectively refreshing the tracking accuracy parameter.

Inventive Principle:

Principle #35Parameter changes

2Speed

If continuous pose estimation is performed, then real-time tracking is achieved, but computational resources are continuously consumed

Engineering Contradiction:

Improvetracking speedVSAvoidcomputational energy

Core Design Contradiction:

SpeedVSUse of energy by moving object

Solution Approach 1:

The patent implements periodic action by performing full pose correction only when reference objects are detected, rather than continuously. The system operates in a cycle of continuous lightweight tracking followed by periodic correction events triggered by reference object detection. This reduces computational energy consumption while maintaining tracking speed, as the intensive correction calculations are performed only when necessary.

Inventive Principle:

Principle #19Periodic action

3Measurement precision

If reference objects are used for pose correction, then pose drift is reduced, but the system requires detectable reference objects in the environment

Engineering Contradiction:

Improvepose accuracyVSAvoidenvironment adaptability

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent enhances universality by designing the system to handle multiple scenarios: when reference objects are detected, it performs correction-based tracking; when reference objects are not detected, it continues with estimation-based tracking. This multi-functional approach allows the system to adapt to different environmental conditions, maintaining operational versatility while achieving high accuracy when reference objects are available.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS12223676B2Method for pose correction and host

Publication Date: 2025.02.11 HTC CORP

AI summary

The embodiments of the disclosure provide a method for pose correction and a host. The method includes: obtaining a first image; in response to determining that a first reference object of at least one reference object exists in the first image, determining a first relative position between the host and the first reference object; obtaining a first reference pose based on the first relative position; and correcting a pose of the host based on the first reference pose.