Method and system for iterative motion inference and calibration using multi-sensor systems

Abstract

The present disclosure provides a system for motion inference and calibration, comprising a network of IMUs aligned in fixed relative positions, and a method to calculate body dimensions based on arcs of movement and known distances between sensors.

Description

METHOD AND SYSTEM FOR ITERATIVE MOTION INFERENCE AND CALIBRATION USING MULTI-SENSOR SYSTEMSBACKGROUND

[0001] 1. Technical Field

[0002] The present disclosure pertains to systems and methods for motion inference and body mapping through iterative calibration using multiple sensors, such as IMUs, force sensors, and biometric inputs.

[0003] Applications include sports performance tracking, humanoid robotics, exosuits, and wearable assistive devices.

[0004] 2. Description of the Related Art

[0005] Accurate motion tracking traditionally requires external equipment such as cameras or motion capture systems, which are costly and location-dependent. Existing wearable solutions often fail to provide detailed motion inference or require manual calibration. There is a need for a system that can dynamically infer dimensions and motion using sensors, adapt over time, and work independently of external setups.SUMMARY

[0006] This invention introduces a method and system for iterative motion inference and calibration using sensor networks. By integrating data from multiple IMUs and other sensors, the system calculates body dimensions, joint positions, and motion patterns without external references.

[0007] Key innovations include:

[0008] 1. Arc and Radius-Based Inference: Use of arcs generated by sensor movement to calculate distances between sensors and body joints.

[0009] 2. Dimensional Cross-Referencing: Layered verification of derived dimensions through biometric and environmental inputs.

[0010] 3. Iterative Refinement: Machine learning algorithms refine the model by comparinginferred dimensions with known datasets.

[0011] 4. Hierarchical Verification: Cross-referencing motion data with stable outputs for accuracy testing.

[0012] 5. Dynamic Adaptation: Algorithms that improve accuracy over time using passive learning from user actions.

[0013] Applications extend to humanoid robotics, where systems can map their own dimensions and range of movement, and exosuits, which can adapt to individual users without manual calibration.

[0014] In an embodiment, the present disclosure provides a system for motion inference and calibration, comprising: a network of IMUs aligned in fixed relative positions; a method to calculate body dimensions based on arcs of movement and known distances between sensors; cross-referencing inferred dimensions with biometric and environmental inputs.

[0015] In an embodiment, the present disclosure provides a method for determining joint locations and body dimensions using deviations in arcs and radii, enabling multi-joint mapping and iterative refinement.

[0016] In an embodiment, the present disclosure provides a system for refining motion models, integrating: biometric data layers for enhanced verification; prior datasets for pattern comparison; machine learning algorithms for real-time model updates.

[0017] In an embodiment, the present disclosure provides a dynamic motion tracking framework, enabling: motion inference using paired or pyramid-aligned IMUs for single-plane or 3D tracking; continuous adaptation through iterative updates based on new inputs.

[0018] In an embodiment, the present disclosure provides applications of the system, including: sports performance tracking; humanoid robotics with self-mapping capabilities; exosuits for assistive motion.

[0019] The technical objectives to be achieved by the disclosure are not limited to those described herein, and other technical objectives that are not mentioned herein would be clearly understood by a person skilled in the art from the description of the disclosure.BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The above and other aspects, features, and advantages of certain embodiments of thedisclosure will be more apparent from the following description taken in conjunction with the accompanying drawings, in which:

[0021] FIG. 1 illustrates a sensor placement on an arm in an embodiment;

[0022] FIG. 2 illustrates detecting orientation using gravity in an embodiment;

[0023] FIG. 3 illustrates a shift compensation in an embodiment;

[0024] FIG. 4 illustrates a new generation of a sensor with EMG and a force sensor in an embodiment;

[0025] FIG. 5 illustrates a body mapping diagram in an embodiment;

[0026] FIG. 6 illustrates a body mapping diagram in an embodiment;

[0027] FIG. 7 illustrates a sport tracking example in an embodiment; and

[0028] FIG. 8 illustrates a humanoid introspection example in an embodiment.DESCRIPTION OF THE EMBODIMENTS

[0029] As the disclosure allows for various changes and numerous embodiments, some embodiments will be illustrated in the drawings and described in detail in the written description. However, this is not intended to limit the disclosure to particular modes of practice, and it is to be appreciated that all changes, equivalents, and substitutes that do not depart from the spirit and technical scope of the disclosure are encompassed in the disclosure.

[0030] Unless otherwise defined or implied herein, all terms (including technical and scientific terminology) used in this document retain their commonly understood meanings as recognized by those skilled in the relevant field. Furthermore, it is to be understood that terms, including those defined in standard dictionaries, should be interpreted in a manner that is consistent with their usage within the context of the relevant field and this disclosure. These terms should not be construed in an idealized or overly formal sense unless explicitly specified otherwise herein.

[0031] As is standard practice in the field, certain embodiments are described and depicted in the accompanying drawings using functional blocks, units, components, and / or modules. Those skilled in the art will recognize that these blocks, units, components, and / or modules may be physically realized by electronic (or optical) circuits, including but not limited to logic circuits, discrete elements, microprocessors, hard-wired circuits, memory elements, wiring connections, and similar components. These may be fabricated using semiconductor-based processes or othermanufacturing technologies. In cases where the blocks, units, components, and / or modules are implemented via microprocessors or comparable hardware, they may be programmed and managed using software (e.g., microcode) to execute various functions described herein and may also be driven by firmware and / or software. It is further contemplated that each block, unit, component, and / or module may be implemented using dedicated hardware or a combination of dedicated hardware for specific functions alongside a processor (e.g., one or more programmed microprocessors and associated circuitry) to execute other functions. Additionally, each block, unit, component, and / or module in certain embodiments may be divided into two or more interacting and discrete elements without departing from the scope of this disclosure. Conversely, the blocks, units, components, and / or modules of some embodiments may be integrated into more complex structures without deviating from the scope of the disclosure.

[0032] • Sensor Configuration: IMUs are arranged in fixed relative positions, either in pairs (for single-plane tracking) or pyramidal sets (for multi-plane tracking). Sensors must support high-resolution input and multi-vector detection.

[0033] • Motion Inference Process: By analyzing arcs of movement and deviations, the system calculates body dimensions, compensating for noise or misalignment through layered crossreferencing.

[0034] • Machine Learning Integration: Algorithms refine inferred dimensions over time, leveraging both biometric datasets and motion patterns. Results are tested against stable, well-understood actions.

[0035] • Applications and Embodiments:

[0036] 1. Sports Tracking: Enhance athlete performance by mapping motion dynamics.

[0037] 2. Humanoid Robotics: Robots equipped with sensors self-map body dimensions and build internal motion models. Integration with force sensors allows adaptation to physical resistance.

[0038] 3. Exosuits: Enable rapid adaptation to individual users, eliminating the need for manual calibration.

[0039] Advantages

[0040] 1. Eliminates reliance on external motion capture systems.

[0041] 2. Adapts dynamically to individual users.

[0042] 3. Combines biometric and motion data for unparalleled accuracy.

[0043] 4. Enables real-time refinement of body models.

[0044] 5. Extends functionality to robotics, healthcare, and sports.

[0045] Referring to FIG. 1, FIG. 1 illustrates a sensor placement on an arm in an embodiment. FIG. 1 shows how the sensor blocks may be positioned on a human's arm.

[0046] Referring to FIG. 2, FIG. 2 illustrates detecting orientation using gravity in an embodiment. FIG. 2 shows the coordinates of each sensor block when positioned around the human's arm. The sensor IMU is constantly under the effect of gravity, and thus this information may be used to predict the orientation the armband with respect to the world coordinate system. In addition, through user's movement, the local coordinates (body coordinate) and where the armband is located may be pinpointed, and thus the pose of the human may be better inferred.

[0047] Referring to FIG. 3, FIG. 3 illustrates a shift compensation in an embodiment. FIG. 3 shows an effect of shifting the orientation of the placement of the armband. The orange data and the orange armband show the original placement orientation, and the magenta armband shows a shifted version. In the first case, the shift can simply result in a small phase shift in the data which can be readily measured and compensated. In the second case, when the shift is larger than a certain degree, it can result in both phase shifting and the amplitude degradation. Combining the reading from two (2) adjacent sensors to recover the original orange data may be required.

[0048] Referring to FIG. 4, FIG. 4 illustrates a new generation of a sensor with EMG and a force sensor in an embodiment. To measure muscle activity, the sensor block can be equipped with force sensor and EMG sensor. The EMG sensor may measure the electricity emitted via muscle activity (surface EMG) and in addition, when users contract / relax their muscles, an additional layer of force information can be used to better understand the underlying muscle activity, especially when the muscle that triggers the EMG is not being measured.

[0049] Referring to FIG. 5, FIG. 5 illustrates a body mapping diagram in an embodiment. FIG. 5 shows how simple actions can be used to help the system learn the body kinematic structure of the human. Given the missing information of the human body indices (e.g., how tall are they, how long is their arm span, etc.), the action prompter can prompt the associated list of activities that can be best used to generate the required information. There may be two (2) types of actions: static actions that can be used to measure orientation, and dynamic / rhythmic actions that can be used to study relative distance / length-related information, such as how tall is the armband mounted, how far away from the shoulder joint is the armband mounted, etc.

[0050] Referring to FIG. 6, FIG. 6 illustrates a body mapping diagram in an embodiment. FIG. 6 shows sample actions for both categories.

[0051] Referring to FIG. 7, FIG. 7 illustrates a sport tracking example in an embodiment. FIG. 7 shows how the learned model can be used to infer motion in different coordinates, as well as trajectories of whole body motion. The inferred trajectories can then be used for different applications in sports, such as golf swing analysis, running gait analysis, tennis serve and stroke analysis, baseball swing and throw analysis, etc.

[0052] Referring to FIG. 8, FIG. 8 illustrates a humanoid introspection example in an embodiment. The multi-point trajectory tracking from the model can be used as an additional whole body motion information layer for humanoid robot to capture nuances and accumulated errors for downstream control systems.

[0053] The above description is an example of technical features of the disclosure, and those skilled in the art to which the disclosure pertains will be able to make various modifications and variations. Thus, the embodiments of the disclosure described above may be implemented separately or in combination with each other.

[0054] The embodiments disclosed in the disclosure are intended not to limit the technical spirit of the disclosure but to describe the technical spirit of the disclosure, and the scope of the technical spirit of the disclosure is not limited by these embodiments. The protection scope of the disclosure should be interpreted by the following claims, and it should be interpreted that all technical spirits within the equivalent scope are included in the scope of the disclosure.

[0055] SUPPLEMENT: Extended Applications and Embodiments

[0056] Advanced Robotics Data Collection and Teleoperation: While the system provides wholebody motion information to capture nuances and accumulated errors for downstream control systems, it is further configured for direct, real-time teleoperation. The inferred multi-point trajectories and muscle-force readings are mapped to robotic kinematic chains. This enables high-fidelity human-in-the-loop data collection and model training for autonomous systems.Embodiments include mapping human arm and hand movements to bio-manipulators, robotic arms, and humanoid arms; and translating human limb movements into locomotor commands for legged robots.

[0057] Immersive Gaming and Virtual Utilities: The system translates localized physical movements and force-sensor data into digital environments. Users can map their specific bodilymotions to dictate the skills, mechanics, or movements of an in-game avatar. For example, specific muscle contractions or arm trajectories can trigger unique mechanics for controlling characters, providing a deeply integrated, controller-free AR / VR and PC gaming experience.

[0058] Advanced Sports Analytics and Custom Biomechanical Planning: Expanding upon standard sports performance tracking, the system measures multi-dimensional athlete movements to generate personalized biomechanical insights. By using iterative refinement algorithms, the system generates tailored improvement plans, such as optimizing a running plan for gait efficiency or correcting weight-training form, to maximize physical outcomes and minimize injury risk.

[0059] Sign Language Decoding and Gesture Translation: The integration of the IMU network with surface EMG and force sensors allows for the highly accurate capture of complex hand, wrist, and arm movements. The system analyzes the electricity emitted via muscle activity alongside force information to decode American Sign Language (ASL) and other gesture-based communications in real-time, translating physical signs into text or synthetic speech.

[0060] Occupational Ergonomics and Remote Rehabilitation: The system's dynamic motion tracking framework operates as a continuous monitor for physical therapy patients or workers. It quantifies joint mobility recovery over time or detects ergonomic degradation in occupational settings, automatically alerting users to postural deviations.

Claims

WHAT IS CLAIMED IS:

1. A system for motion inference and calibration, comprising:a network of IMUs aligned in fixed relative positions;a method to calculate body dimensions based on arcs of movement and known distances between sensors;cross-referencing inferred dimensions with biometric and environmental inputs.

2. A method for determining joint locations and body dimensions using deviations in arcs and radii, enabling multi -joint mapping and iterative refinement.

3. A system for refining motion models, integrating:biometric data layers for enhanced verification;prior datasets for pattern comparison;machine learning algorithms for real-time model updates.

4. A dynamic motion tracking framework, enabling:motion inference using paired or pyramid-aligned IMUs for single-plane or 3D tracking; continuous adaptation through iterative updates based on new inputs.

5. Applications of the system, including:sports performance tracking;humanoid robotics with self-mapping capabilities;exosuits for assistive motion.

6. The system of claim 1, further configured for robotic teleoperation and data collection, wherein the inferred body dimensions and multi-point trajectories are mapped to robotic kinematic chains to control humanoid bio-manipulators, robotic arms, and legged robots for model training and real-time operation.

7. The system of claim 1, further comprising a gaming and virtual utility interface, wherein the inferred motions, muscle activity, and force data are translated into character skills and mechanics within digital, PC, and AR / VR environments.

8. The system of claim 1, further comprising an analytics engine for sports andphysical fitness, configured to measure athletic movements, compare them against optimal biomechanical datasets, and automatically generate custom movement improvement plans, running schedules, and form-correction protocols.

9. A method for decoding sign language and complex gestures, leveraging the system of claim 1, comprising: capturing spatial trajectories via the IMU network; combining trajectory data with underlying muscle activity measured via EMG and force sensors; translating the combined sensor data into decoded text or audio outputs representing specific signs or language elements.

10. The system of claim 1, further applied to clinical rehabilitation and occupational safety, configured to passively track joint mobility, measure range-of-motion improvements, and detect postural anomalies to prevent repetitive strain.

Images