Learned Model Swing Direction Detection in Game Controllers

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing game processing systems using inertial sensors struggle to accurately determine the direction of a swing input, leading to inconsistencies in game responses.

Innovation Solution

A computer-readable storage medium with a game program that uses operation data from an inertial sensor-equipped operation device to determine whether a swing input has been performed and to identify the swing direction using a learned model generated from teacher data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional swing detection methods are used, then the device complexity remains low, but the measurement precision of swing direction is insufficient

Engineering Contradiction:
Improveswing direction determination accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The learned model is generated in advance through offline learning using teacher data collected from multiple users. This pre-computed model is then stored and directly applied during actual game operation, eliminating the need for real-time complex calculations while achieving high swing direction determination accuracy

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

A learned model acts as an intermediary between the raw acceleration data from the inertial sensor and the final swing direction determination. The model processes the sensor data and outputs swing direction information, simplifying the overall system while improving measurement precision

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If swing determination is performed quickly, then the productivity is high, but the measurement precision of swing direction may be compromised

Engineering Contradiction:
Improveswing direction determination accuracyVSAvoidprocessing speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The learned model performs swing direction determination in real-time during game operation, providing quick results. The complex learning and data processing were performed in advance offline, allowing fast online execution without compromising accuracy

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent replaces complex real-time mechanical analysis of swing motion with a pre-trained learned model that processes acceleration data through machine learning algorithms, achieving both speed and accuracy in swing direction determination

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS12233336B2Computer-readable non-transitory storage medium having game program stored therein, game system, game apparatus, and game processing method
Publication Date: 2025.02.25 NINTENDO CO LTD
  • US12233336B2 patent drawing
  • US12233336B2 patent drawing
  • US12233336B2 patent drawing

AI summary

On the basis of operation data acquired from an operation device including an inertial sensor, whether or not a swing input to the operation device has been performed is determined. A learned model generated on the basis of a plurality of teacher data each associated with one of a plurality of swing directions, the learned model being for determining which swing direction out of the plurality of swing directions the operation device has been swung in, is managed. The operation data acquired in a period in which the swing input has been performed is inputted to the learned model, and a swing direction in which the operation device has been swung is determined on the basis of an output from the learned model corresponding to the input. Game processing is executed on the basis of the swing direction.