Systems and methods for personalized motion control

Abstract

End users, unskilled in the art, generating motion recognizers from example motions, without substantial programming, without limitation to any fixed set of well-known gestures, and without limitation to motions that occur substantially in a plane, or are substantially predefined in scope. From example motions for each class of motion to be recognized, a system automatically generates motion recognizers using machine learning techniques. Those motion recognizers can be incorporated into an end-user application, with the effect that when a user of the application supplies a motion, those motion recognizers will recognize the motion as an example of one of the known classes of motion. Motion recognizers can be incorporated into an end-user application; tuned to improve recognition rates for subsequent motions to allow end-users to add new example motions.

Description

CROSS REFERENCE TO RELATED APPLICATIONS[0001]This is a continuation-in-part of co-pending U.S. application Ser. No. 11 / 486,997, entitled “Generating Motion Recognizers for Arbitrary Motions”, filed Jul. 14, 2006, and co-pending U.S. application Ser. No. 12 / 020,431, entitled “Self-Contained Inertial Navigation System for Interactive Control Using Movable Controllers”, filed Jan. 25, 2008.BACKGROUND OF THE INVENTION[0002]1. Field of the Invention[0003]The invention generally relates to the area of artificial intelligence, and more particularly, relates to machine learning, especially in the context of generating motion recognizers from example motions. In some embodiments, recognizer makers can be incorporated into, or used alongside of end-user applications, where end users can create ad-hoc personalized motion recognizers for use with those end-user applications.[0004]2. Related Art[0005]Our ability to fulfill the promise of freeform human motion control of software applications is ...

AI Technical Summary

Benefits of technology

[0053]receiving a second set of motion signals from a second motion sensing device providing sufficient information to compute position and orientation over time of the second motion sensing device; and

Problems solved by technology

Our ability to fulfill the promise of freeform human motion control of software applications is strictly limited by our ability to detect and recognize what a given human is trying to do.

Humans are very good at interpreting the gestures and expressions of other humans, but are yet unable to create machines or code that can perform at a similar level.

Writing program code to recognize whether a supplied motion is an example of an existing set of known motion classes is difficult.

The resulting motion data is often complicated and counterintuitive.

For example, when presented with a simple graph of 3D accelerometer outputs versus time, people skilled in the art struggle to determine what gesture that time series of data corresponds to.

Even the simpler task of selecting which motion graphs belong to the same gesture confounds most experts presented with the problem.

The problem is exacerbated by sensor noise, device differences, and the fact that data for the same gesture can appear quite different when performed by different people with different body types and musculatures, or even by the same person at different times. It is a difficult challenge under these conditions for one skilled in the art to build effective motion recognizers.

Along with challenging source data, the fact that the data is dynamic over time, not static over time, is a significant hurdle to overcome.

To our knowledge, there is nothing in the prior art that provides teaching related to end-user creation of ad-hoc motion recognizers.

This approach does not work in applications that wish to provide freeform motion control, since the starting and ending positions are not predefined, and can not reasonably be quantized a priori without making the construction of a reasonable training set a virtual impossibility.

Images