patent.fyi
Applications  /  US 2026/0355573
United States Patent Application Publication

US 2026/0355573 A1

Published
Nov 12, 2026
Filed
Sep 11, 2023
Earliest date on the face
Sep 11, 2023
Classification
A63B 24/0003
First published here
Nov 19, 2026

Generated automatically. Not reviewed by a person. Not legal advice or an opinion.

Section 1 · machine-generated

What if?

Questions about how individual parts of claim 1 might be changed. A question may describe something different from what the claim language recites, or a version of it; no view is offered on which.

1.1Manual entry of joint angles

What ifWhat if a coach entered observed joint angles by hand as the motion input?
Claim language“a sensor configured to acquire motion data”

A coach watches three repetitions and enters a joint angle to the nearest five degrees at the start, midpoint, and end of each repetition. The system assembles the entries in phase order and calculates the range of motion for each repetition.

1.2Fixed digital arithmetic circuits

What ifWhat if fixed digital circuits implemented the computation without storing processing instructions?
Claim language“memory storing processing instructions”

A dedicated digital circuit routes digitized joint angles through fixed comparator, multiplier, and adder stages. Registers hold samples and intermediate values, and hardwired control logic advances the calculation at each sample clock pulse.

1.3Continuous analog prediction circuit

What ifWhat if a continuous analog circuit performed the calculation without executing processing instructions?
Claim language“processor configured to execute the processing instructions”

An active low-pass filter smooths the joint-angle voltage, and a peak-hold circuit captures its maximum during each repetition. A summing amplifier combines that maximum with an exercise-load voltage using resistor ratios that set the contribution of each input.

1.4Direct prediction of sensor samples

What ifWhat if the system forecast future sensor readings directly from the incoming samples?
Claim language“extract motion feature data”

The system places the latest 100 unmodified samples from each sensor axis in a rolling buffer. A multichannel autoregressive model takes the buffered samples and the exercise load as inputs and forecasts the next 20 samples on each axis.

1.5Forecast from motion history

What ifWhat if the predictor used only changes observed in earlier repetitions?
Claim language“based on exercise data representing content of exercise performed”

A rolling buffer holds the peak knee flexion measured over the last eight repetitions. A least-squares fit estimates the change in degrees per repetition and projects the peak for the next repetition.

Section 2 · machine-generated

Variations

Further arrangements of the same subject matter, described here as of the publication date at the foot of this page.

Other field of use

Reach rehearsal at an assembly station

A depth camera mounted above an assembly bench records a worker performing a practice reach between two fixtures. A processor extracts peak shoulder elevation from the recording and reads the reach direction, fixture height, and handled mass from a practice log. A regression model stored in memory predicts the change in peak shoulder elevation after the logged practice, and a renderer uses the predicted angle to set the apex of a reach animation displayed beside a model of the bench.

Component substitution

Foot sensing with inertial measurement units

Two shoe-mounted six-axis inertial measurement units replace the camera as the motion sensor and transmit timestamped acceleration and angular velocity samples to a processor. The processor detects heel strikes and calculates stride period from successive strikes of the same foot. A model stored in memory predicts the change in stride period from the measured period and the treadmill speed and interval duration logged during training. The renderer retimes a running animation to the predicted stride period and encodes the resulting frames as video.

State transition model for squat depth

A side-view camera records a squat, and a processor extracts the minimum hip height relative to standing hip height. A linear state transition model replaces the machine learning predictor and uses the measured height ratio, completed repetitions, and lifted mass to calculate the change in the ratio over the next exercise block. The processor clamps the resulting ratio to the animation rig's supported range and uses it to set the hip height at the bottom of the rendered squat.

Parameter or scale variation

High speed recording for jump landings

A side-view camera captures a two-second landing sequence at 240 frames per second and stores the frames in a circular buffer. The processor measures peak knee flexion during the first 150 milliseconds after ground contact and predicts its change from the measured angle and the completed drop height and repetition count. The renderer applies the predicted angle to a landing animation and outputs 120 frames at 60 frames per second to display half a second of predicted motion at one-quarter speed.

Combination with existing systems

Cycling forecast on a trainer display

A side-view camera connects to the console of an electronically braked cycling trainer, which logs completed intervals with their resistance settings and durations. The console processor estimates knee extension at the bottom of the pedal stroke and feeds that angle and the interval log to a stored prediction model. The console adjusts the saddle-to-pedal animation rig to show the predicted knee extension and displays the resulting pedaling video next to the interval summary.

Automated or data-driven variant

Automatic exercise logging for rowing forecasts

A camera captures a side view of the rower, while a handle position encoder and a tension sensor supply synchronized measurements to the processor. Software detects stroke boundaries at handle direction reversals and logs completed stroke count and mean handle tension for each exercise block. A trained regression model combines the automatically logged exercise data with the torso angle measured from the camera at the finish of the stroke to predict the change in that angle. A renderer sets the finish angle of a rowing animation to the predicted value and outputs the frames as video.

Specific version of a part

Shank mounted inertial sensor

An elastic strap secures a three-axis accelerometer and a three-axis gyroscope to the front of the shank. The sensor records acceleration and angular velocity at 100 samples per second and attaches a timestamp to each sample.

Peak flexion per repetition

The system subtracts the sagittal-plane orientation of the shank from that of the thigh to calculate knee flexion. An upward crossing of 20 degrees starts a repetition, and a downward crossing of the same threshold ends it. The system records the maximum angle between the two crossings.

Regression with exercise load

A ridge regression model accepts the exercise load in kilograms, the number of completed repetitions, and the mean peak knee flexion over the last three repetitions. It outputs the predicted change in peak flexion in degrees for the next repetition.

Side view stick figure video

The renderer interpolates the predicted hip and knee angles between exercise phases and applies them to a stick figure with fixed segment lengths. It renders the figure from the side at 30 frames per second and encodes the frames for playback.

Section 3 · quoted from USPTO records

Other work

Passages reproduced word for word from two earlier-published U.S. patent documents, placed beside the language of claim 1. No characterization is offered. Each passage is marked A or B for the document it comes from; ¶ numbers are the paragraph numbers in the USPTO’s published full text of that document.

Document AUS 2017/0128783 A1
Published
May 11, 2017
Filed
Mar 23, 2015
Class
A63B 24/0075
Document BUS 2019/0083847 A1
Published
Mar 21, 2019
Filed
Jul 12, 2018
Class
A63B 24/0087
Claim 1
Document A
Document B
1.1a sensor configured to acquire motion data representing a physical motion of a person;
In this embodiment, the detecting section 22 includes a pulse wave sensor that detects a pulse wave as the biological information and an acceleration sensor that detects, as the motion information, an acceleration value acting according to the motion of the user.
A · US 2017/0128783 A1 · ¶ 47
For example, the movement sensing device may include a hardware such as camera or the like, and be configured to detect the body movements of the exerciser by means of motion capturing techniques.
B · US 2019/0083847 A1 · ¶ 90
1.2at least one memory storing processing instructions; and
The storing section 35 is configured by a HDD (Hard Disk Drive), a flash memory, and the like and stores a program and data necessary for the operation of the information processing apparatus 3.
A · US 2017/0128783 A1 · ¶ 61
a control device including: a memory; a processor; one or more modules stored in the memory and configured to be executed by the processor, the one or more modules including instructions: to acquire exercise parameters of the exerciser in outdoor exercises according to a mapping relationship between the body movements of the exerciser and the exercise parameters of the exerciser in the outdoor exercises; and to drive the exercise device to operate according to the exercise pa
B · US 2019/0083847 A1 · ¶ 35
1.3at least one processor configured to execute the processing instructions, wherein the at least one processor is configured to execute the processing instructions to:
The control section 37 includes a CPU (Central Processing Unit). The CPU executes the program stored in the storing section 35, whereby the control section 37 controls the operation of the information processing apparatus 3.
A · US 2017/0128783 A1 · ¶ 66
1.4extract motion feature data representing a feature of a motion from the motion data acquired by the sensor;
The movement sensing device 301 may detect the body movements of the exerciser by means of motion capture techniques.
B · US 2019/0083847 A1 · ¶ 140
1.5predict predicted motion feature data representing a change in the motion feature data based on exercise data representing content of exercise performed by a the person and on the motion feature data; and
1.6generate, based on the predicted motion feature data, predicted video data for displaying a predicted physical motion of the person.

An empty cell means only that no passage was reproduced there.

Statements in these documents
Because of such a background, there has been a demand for a configuration that can present exercise corresponding to a purpose.
A · US 2017/0128783 A1 · ¶ 7
One of objects of the present invention is to provide an exercise presenting apparatus, an exercise presenting method, and an exercise presenting program that can present effective exercise.
A · US 2017/0128783 A1 · ¶ 8
On the one hand, exercises on exercise devices such as treadmills and the like are not as healthy and comfortable as outdoor exercises.
B · US 2019/0083847 A1 · ¶ 4
Unfortunately, exercises on the exercise devices such as treadmills may disrupt an exercise rhythm of the exerciser, resulting in that the exerciser is unable to keep his/her outdoor exercise habit.
B · US 2019/0083847 A1 · ¶ 4
Cite aspatent.fyi, US 2026/0355573 A1, claim 1, first published 2026-11-19T13:13:53-05:00, https://patent.fyi/US20260355573/, SHA-256 7cb68660c82d972f6c3134f1b0f37533d759fe77c972fa6c4107c5e6df5531b1
First published
Permanent addresshttps://patent.fyi/US20260355573/
Content fingerprint (SHA-256)7cb68660c82d972f6c3134f1b0f37533d759fe77c972fa6c4107c5e6df5531b1
Methodv2026-11-18.2 · automated, not reviewed by a person
Correctionscorrections@patent.fyi