Machine Learning Classification of Real-World Points of Interest

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing parallel reality games lack sufficient data on real-world points of interest, which hinders the integration of rich game features and interactions, limiting the connection between virtual and real-world experiences.

Innovation Solution

A machine learning model is employed to classify points of interest using training data sets, incorporating player inputs to verify and generate properties for these locations, enabling the game server to determine suitability for various activities and enhance game features.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a parallel reality game uses basic location data only, then the game can be implemented with simple data requirements, but the game features and interactions remain limited and lack richness

Engineering Contradiction:

Improvegame featuresVSAvoiddata requirements

Core Design Contradiction:

Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The system performs preliminary classification of points of interest using a machine learning model before game features are implemented. Training datasets are prepared in advance with verified properties, and the model is trained beforehand to automatically classify POIs, so that when game features need to be added, the classification data is already available to support diverse interactions.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The machine learning model automatically classifies points of interest and generates properties without requiring manual verification for each POI. The system uses player inputs as training data that continuously improve the model's classification accuracy, creating a self-improving system where the game itself generates the data needed for rich features.

Inventive Principle:

Principle #25Self-service

2Measurement precision

If manual verification of point of interest properties is performed, then data accuracy is improved, but the time and resources required increase significantly

Engineering Contradiction:

Improveproperty accuracyVSAvoidverification time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The manual mechanical verification process is replaced with an automated machine learning classification system. The ML model processes point of interest data automatically, using training datasets to verify properties and generate classifications without human intervention, thereby maintaining high accuracy while eliminating the time cost of manual verification.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system uses player inputs as feedback to continuously improve the classification model. Player-verified properties are fed back into the training dataset, allowing the model to learn from real-world corrections and improve its accuracy over time, creating a self-correcting verification system.

Inventive Principle:

Principle #23Feedback

3Adaptability or versatility

If rich property data is collected for all points of interest, then the suitability for various game activities can be determined, but the complexity of data collection and processing increases

Engineering Contradiction:

Improveactivity suitabilityVSAvoiddata processing system

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The data collection and processing system is segmented into modular components: a training data collection module, a machine learning model training module, and a classification inference module. Each component handles a specific aspect of the data pipeline, making the overall complex system manageable and maintainable through clear separation of concerns.

Inventive Principle:

Principle #1Segmentation

4Reliability

If player inputs are used to verify properties, then data accuracy is improved through community validation, but the system requires coordination and processing of multiple user inputs

Engineering Contradiction:

Improveproperty verificationVSAvoidinput coordination system

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

Multiple player inputs are merged into a unified training dataset that feeds the machine learning model. The system combines corrections, verifications, and additions from multiple players into a single coherent data structure, eliminating the need for complex individual verification workflows while maintaining the benefits of community validation.

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS12447406B2Classifying real-world points of interest in a parallel virtual world

Publication Date: 2025.10.21 NIANTIC SPATIAL INC

AI summary

A machine learning model classifies points of interest in a parallel reality game hosted by a server. The server generates training data sets that include verified properties for points of interest. The machine learning model may predict unverified properties for points of interest. Players in the parallel reality game may input properties for the points of interest. The machine learning model use the received properties from players as inputs to the machine learning model to verify unverified properties or generate new properties for the points of interest. The server may classify the points of interest as suitable for particular activities, and the server may use the classifications for future activities within the parallel reality game.