DESC:FIELD OF THE INVENTION

The present invention embodies a system and method for enabling skill transfer through a human-to-human interface. More specifically, the invention provides a system and method for training a trainee by transferring skills from multiple experts using a human-to-human interface—whether in online or offline mode—while utilizing surface EMG (sEMG) signals integrated with advanced deep learning features to enhance the efficacy of skill acquisition and training.

BACKGROUND OF THE INVENTION

Learning or Intelligence is not hard coded in any individual’s DNA. It is years of hard work, strenuous efforts and immense time one invests in a trade to secure a badge of proficiency and accomplishment. In humans, brain’s cortical plasticity helps it to assist and repair it’s cognitive and motor impairments and learn new skills. However, cognitive brain being the most incredibly complex and composite organ of human body is still not fully decoded for its deep compartmentalized and yet intertwined encodings. It will be a matter of another decade or so, when humans can claim successful deciphering of brain-body interactions.

Nevertheless, with recent milestone advances in the field of neuroscience, functioning of complex central nervous system involving a network of 80 billion neurons for continuous transmission and receiving of chemical and electrical messages has helped researchers in assisting human body in repair and maintaining all bodily functions besides enabling humans to learn and develop new skills and memory based on past experiences. Based on this research, various brain invasive, non-invasive or partially invasive tools and instruments have been devised to understand brain regions that get stimulated as brain learns and conditions itself with new environment and novel experiences.
As the muscles in human body perform repetition of any action and form conditioned reflexes to an external/internal stimuli, the muscle memory is build that gets stored in brain and hardwired with years of practice. This is how one acquires expertise by repeating certain sequence of actions that reinforces neural pathways associated with specific movements, making them more automatic and less time and effort intensive. Off course, even thinking of starting learning and developing a new skill is a major deterrent for many of us. Once neural pathways are formed, it is difficult to subject them to new challenges of re-wiring as one progressed in age. That is the reason it is easier to learn new skills when we are young compared to older generations.

Now, what if all these years of accumulated experience and toiling labour is made transferrable from an expert to an amateur that can minimize time for learning and increase effectiveness of human learning experience. With highly effective human to human interfacing devices studied and developed with great interest in last decade, this transfer of acquired learning via feeding of neural impulses acquired from expert to a trainee/devout/novice to help the latter build same type of muscle memory as that of an expert has been made possible, though to a very restricted scale and scope.

In many notable endeavours, various invasive/non-invasive techniques have been used to capture brain and other motor signals, with Electroencephalography (EEG) being the most commonly deployed technique. The simple technique has been capturing of EEG signals from motor actions using electrodes and converting captured neural impulses to electrical signals. These electrical signals are assessed for their frequency, amplitude, pulse width and the like so that any converter may be used to adjusting these electrical current pulses to frequency and pulse width of the trainee’s neuro-muscular system. This helps the trainee perform same motor function like that of an expert in a real-time scenario.

This technique can be highly valuable in skill training across all spheres including healthcare, automation, defence, sports, navigation and the like. In fact, the approach can even be manifested on the automatic robots that can be trained to perform tasks similar to humans instead of getting into the loop of complex coding and associated malfunctions and failures. Humans, on the other hand can upskill themselves from the experts of the field with utmost precision and accuracy. It will be also easier to train human for multi-tasking as the less used hand/feet can be trained via human to human interface to perform less mentally engaging and more mundane and repetitive tasks, helping them save time and attain higher physical and spiritual limits.

However, the research so far has only been able to achieve success on live capturing of electrical pulses of expert for innervating the trainee target muscle group in real time. This restricts the transfer of skill set from only one expert at a time, and that too accompanied with basic human limitations. Further, it mandates the presence of expert at all times for transferring the instructions to the trainee, which obviously impacts the scalability of any such solution. Furthermore, the highest feat achieved by any individual in any domain may still be challenged for further improvisation and up gradation for reasons of impediments in human capabilities and other practical limitations.

Nevertheless, as a trainee one would prefer to have his learning and training from not just one but several maestros of the trade so that he can pick the best of each. Finding such a unique combination of varied skills in one sole individual is unrealistic and too far-fetched. Also, using presently available human to human interfacing technique has only helped so far with real time muscle stimulation of a trainee from a single expert at any instance. However, demand is ever for more and one to be met at all times.

In this vein, the present disclosure sets forth system and method for enabling skill transfer from multiple experts of trade to a trainee/devout via a human to human interface; the system and method embodying advantageous alternatives and improvements to assist trainee in developing muscle memory as specific neural impulses generated at trainee’s end are processed to learn corresponding motor action and transferred as feed to trainee at any point of time, and that may address one or more of the challenges or needs mentioned herein, as well as provide other benefits and advantages.

OBJECT OF THE INVENTION

An object of the present invention is to provide a system and method for enabling accelerated skill transfer from multiple trainers to a trainee via a human to human interface (HHI).

Another object of the present invention is to provide an efficient and time saving system and method for rapid skill transfer by way of developing muscle memory within the trainee from neural impulses of trainers.

Yet another object of the present invention is to provide an effective system and method of duplicating motor actions of one or more trainers within a trainee in both online/offline mode.

Yet another object of the present invention is to provide enhanced and a very quick learning experience by reducing time to develop muscle memory via a human to human interface that captures neural impulses from trainer for transfer to trainee.
Yet another object of the present invention is to provide a more robust, reproducible and repeatable system and method for extracting muscular activity of multiple trainers and classifying the activity using machine learning for recording the physiological patterns and parameters thereof.

In yet another embodiment, the system and method enable transferring motor skills from trainer/expert to trainee/learner mediated by wearable technology.

BRIEF DESCRIPTION OF THE DRAWINGS

So that the manner in which the above recited features of the present invention can be understood in detail, a more particular to the description of the invention, briefly summarized above, may be had by reference to embodiments, some of which are illustrated in the appended drawings. It is to be noted, however, that the appended drawings illustrate only typical embodiments of this invention and are therefore not to be considered limiting of its scope, the invention may admit to other equally effective embodiments.
These and other features, benefits and advantages of the present invention will become apparent by reference to the following text figure, with like reference numbers referring to like structures across the views, wherein:

Fig. 1 illustrates a block diagram for skill transfer via human to human interface, in accordance with an embodiment of the present invention.

SUMMARY
According to the first aspect of disclosure, a method for enabling real-time skill transfer between a trainer and trainee is disclosed, wherein the method comprises of: acquiring electrophysiological signals from one or more trainer(s) via a set of non-invasive, high-sensitivity electrodes embedded within a head mounted device; pre-processing the acquired electrophysiological signals and analysing the pre-processed electrophysiological signals using one or more machine learning models to estimate motor intent, muscle force, identify physiological patterns and extract one or more neuromuscular parameters. The analysed electrophysiological signals are classified into a corresponding muscle activity label; and a neurostimulation protocol is triggered in the trainee based on the classified muscle activity, wherein the stimulation is configured to replicate trainer’s muscle activation pattern in the trainee.

In another aspect of the disclosure, the system for enabling real-time skill transfer between a trainer and trainee, is disclosed, wherein the system comprises: a head mounted device worn by the trainer and the trainee, and configured to acquire electrophysiological signals from one or more trainer(s) via a set of non-invasive, high-sensitivity electrodes embedded within the head mounted device. The system further comprises of a processing module configured to: pre-process the acquired electrophysiological signals and analyse the pre-processed electrophysiological signals using one or more machine learning models to estimate motor intent, muscle force, identify physiological patterns and extract one or more neuromuscular parameters; classify the analysed electrophysiological signals into a corresponding muscle activity label; and
trigger a neurostimulation protocol in the trainee based on the classified muscle activity, wherein the stimulation is configured to replicate trainer’s muscle activation pattern in the trainee.

DETAILED DESCRIPTION
While the present invention is described herein by way of example using embodiments and illustrative drawings, those skilled in the art will recognize that the invention is not limited to the embodiments of drawing or drawings described and are not intended to represent the scale of the various components. Further, some components that may form a part of the invention may not be illustrated in certain figures, for ease of illustration, and such omissions do not limit the embodiments outlined in any way. It should be understood that the drawings and detailed description thereto are not intended to limit the invention to the particular form disclosed, but on the contrary, the invention is to cover all modifications, equivalents, and alternatives falling within the scope of the present invention as defined by the appended claims.

As used throughout this description, the word "may" be used in a permissive sense (i.e., meaning having the potential to), rather than the mandatory sense, (i.e., meaning must). Further, the words "a" or "an" mean "at least one” and the word “plurality” means “one or more” unless otherwise mentioned. Furthermore, the terminology and phraseology used herein is solely used for descriptive purposes and should not be construed as limiting in scope. Language such as "including," "comprising," "having," "containing," or "involving," and variations thereof, is intended to be broad and encompass the subject matter listed thereafter, equivalents, and additional subject matter not recited, and is not intended to exclude other additives, components, integers or steps.

Likewise, the term "comprising" is considered synonymous with the terms "including" or "containing" for applicable legal purposes. Any discussion of documents, acts, materials, devices, articles, and the like are included in the specification solely for the purpose of providing a context for the present invention. It is not suggested or represented that any or all of these matters form part of the prior art base or were common general knowledge in the field relevant to the present invention.

In this disclosure, whenever a composition or an element or a group of elements is preceded with the transitional phrase “comprising”, it is understood that we also contemplate the same composition, element or group of elements with transitional phrases “consisting of”, “consisting”, “selected from the group of consisting of, “including”, or “is” preceding the recitation of the composition, element or group of elements and vice versa.

The present invention is described hereinafter by various embodiments with reference to the accompanying drawings, wherein reference numerals used in the accompanying drawing correspond to the like elements throughout the description. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiment set forth herein. Rather, the embodiment is provided so that this disclosure will be thorough and complete and will fully convey the scope of the invention to those skilled in the art.

In the following detailed description, numeric values and ranges are provided for various aspects of the implementations described. These values and ranges are to be treated as examples only and are not intended to limit the scope of the claims. In addition, a number of materials are identified as suitable for various facets of the implementations. These materials are to be treated as exemplary and are not intended to limit the scope of the invention.

In accordance with one general embodiment of present disclosure, the present system and method are directed to skill transfer from multiple experts/trainers of a trade/occupation to a trainee/devout via a human to human interface by way of developing muscle memory of the trainee. The skill transfer occurs via a human-to-human interface that facilitates the development of muscle memory in the trainee, thereby enabling intuitive and repeatable motor execution of the skill without relying solely on cognitive or visual learning. Apropos, specific neural impulses, particularly surface electromyography (sEMG) signals, generated at trainee’s end during initial attempts or guided sessions are captured, analysed and processed to learn corresponding motor actions involved in performing a particular skill and transferred as feed to trainee in both online/offline mode. These neural and muscular activation patterns are mapped against reference patterns derived from a plurality of expert trainers to ensure consistency and accuracy of the skill being transferred.

The system then generates real-time or batch-mode feedback based on this analysis, which is delivered back to the trainee through haptic, auditory, or visual interfaces to reinforce the correct motor pathways. This feedback loop facilitates neural adaptation and muscular coordination in the trainee over time, resulting in the internalization of the skill at a neuromuscular level. The invention supports both online mode—where the trainee interacts with the system or trainers in real-time—and offline mode, where recorded data, simulations, or asynchronous expert feedback is used to guide the trainee. This dual-mode flexibility allows for scalable deployment across various domains such as industrial training, artisanal crafts, surgical skill development, sports coaching, or rehabilitation therapy.

While the live capturing of trainer’s brain signals from motor actions for mimicking at trainee’s end has been discussed in previous works, this has been strictly limited to live capture of electroencephalograph (EEG) signals from trainer without any capability of storing such signals in offline mode for later training. Also, it is almost impractical to enable live training at all times by the solo trainer as any skill practice would need no less than 6-8 hours each day. This is also improbable in situations where the trainer is remotely located and also when the trainee requires to be trained from more than one trainer to master any skill set.

In background of above limitation, the present disclosure attempts to achieve best in class training for the trainee from one or more trainer’s proficient in distinct techniques, methodologies, or "tricks" of a given trade, craft, or occupational domain. This objective is finely accomplished through an advanced human-to-human skill transfer system 1000, as disclosed herein and shown in Fig. 1, which captures and interprets electrophysiological signals from plurality of expert trainers to facilitate accurate reproduction of neuromuscular patterns in the trainee. In particular, the system 1000 utilizes surface EEG (electroencephalography), surface EMG (electromyography), sensory evoked potentials or other electrophysiological signals to monitor and analyse the trainer’s muscle activation and neural command patterns.

These signals are acquired through a head-worn device 200 or similar wearable instrumentation that is embedded with high-sensitivity electrodes 300 designed to non-invasively detect bioelectrical activity. In one exemplary embodiment, the head mounted device is worn by the trainer is referred by numeral 210 and that worn by the trainee is referred by numeral 220, and collectively referred by numeral 200. Electrophysiological signals may be captured from multiple neuroanatomical loci, including but not limited to:
• the cerebral cortex (responsible for motor planning and execution),
• the brain stem (mediating reflexive and automatic motor functions),
• the spinal cord (conveying neural signals between the brain and peripheral effectors), and
• peripheral nerves (which directly innervate muscles involved in task performance).

Furthermore, the system 1000 is configured to record sensory evoked potentials (SEPs)—which are time-locked neural responses generated by the central nervous system following the stimulation of peripheral sensory organs (e.g., tactile, visual, or proprioceptive stimuli). SEPs serve as a critical indicator of sensorimotor integration and provide an additional layer of insight into how the expert responds to external sensory input during skill execution.

The system 1000 comprises of a processing module 500 that is configured to analyse these electrophysiological recordings using signal processing algorithms and pattern recognition models to extract actionable training data, which is then used to construct feedback loops and adaptive stimuli for the trainee. This enables the encoding and reinforcement of correct motor patterns, effectively leading to the development of muscle memory and neuroplastic changes aligned with expert performance.

The captured electrophysiological signals - such as surface EMG (sEMG) or EEG - are interpreted using machine learning/deep learning algorithm to estimating muscle force, identify physiological patterns and extract other neuromuscular parameters. These interpretations aid in real-time assessment of the user’s motor intent and in predicting muscle activity with high fidelity. However, the raw signals acquired from the electrodes are often susceptible to various forms of noise and artefacts. These may include:
• Transducer noise (e.g., from analog hardware),
• Electrode motion artefacts (caused by skin displacement or poor contact),
• Ambient electrical interference (e.g., powerline noise), and
• biological noise (e.g., from overlapping muscle or cardiac activity).

To ensure the accurate prediction of muscle activation and minimize the degradation caused by a low signal-to-noise ratio (SNR), the captured signals undergo a pre-processing stage involving advanced signal denoising and feature enhancement techniques. In accordance with one example embodiment, the common pre-processing techniques include:

• Band-pass or notch filtering to isolate the relevant frequency band (typically 20–500 Hz for EMG),
• Full-wave rectification to convert the signal into a unidirectional waveform suitable for envelope analysis,
• Wavelet transforms, which enable time-frequency decomposition of the signal for transient feature detection,
• Wave smoothing to reduce signal variability while preserving key features, and
• Fast Fourier Transform (FFT) for spectral feature extraction.

Among these, wavelet transform techniques are often preferred due to their superior time-frequency localization, adaptability to non-stationary signals like EMG, and computational efficiency. Wavelet-based methods can significantly reduce analysis time without compromising the accuracy of feature extraction or classification performance. This makes them particularly suitable for real-time skill transfer systems where both speed and precision are critical.

In one exemplary embodiment, convolutional neural networks (CNNs) or recurrent neural networks (RNNs) including long short-term memory (LSTM) architectures may be employed for assessing change in muscle shape/muscle activation/muscle activity or movement prediction. These models are particularly effective in learning temporal and spatial dependencies from multivariate biosignals. CNNs are useful for recognizing spatial features in EMG signal windows, while LSTMs are adept at modeling time-dependent patterns—such as variations in muscle activation or predicting future motion trajectories based on historical signal data. These models facilitate accurate assessments of muscle shape changes, muscle contraction patterns, and motor intent during dynamic activities.

In one exemplary embodiment, a convolutional neural network (CNN) model is constructed to process various types of biosignals—such as electroencephalography (EEG), electromyography (EMG), surface electromyography (sEMG), or other electrophysiological signals—for the purpose of motion pattern recognition and neuro-motor mapping of actions performed by the trainer. These biosignals serve as rich sources of neuromuscular information and form the basis for decoding motor intent, estimating muscle activity, and facilitating precise skill transfer to the trainee. Specifically, EEG captures the electrical activity of the brain by recording voltage fluctuations along the scalp that result from synchronized ionic currents generated by neurons in the cerebral cortex. EEG is particularly useful for capturing cortical activations associated with motor planning, attention, and decision-making, making it highly relevant for cognitive-motor skill analysis.

On the other hand, EMG—and more specifically surface EMG (sEMG)—records the electrical signals generated by muscle contractions. These signals reflect neural activation patterns at the neuromuscular junction and are indicative of muscle recruitment intensity, coordination, and timing. sEMG provides a non-invasive means of measuring the electrical potential produced by muscle fibers, allowing real-time assessment of musculoskeletal function, muscle fatigue, force generation, and motor control strategies used by skilled trainers.

In accordance with one embodiment, the CNN model is trained on time-series segments or spectrograms of these biosignals to extract spatially and temporally relevant features that correspond to specific motor actions, such as grip strength modulation, limb positioning, or tool manipulation techniques. The convolutional layers help in recognizing localized patterns across time windows, while pooling and fully connected layers enable abstraction and classification of high-level motion classes. This facilitates the construction of a gesture-to-intent map or a skill signature profile for each expert.

In other alternate embodiment, additional or alternative modeling approaches—such as recurrent neural networks (RNNs), long short-term memory (LSTM) networks, transformer-based models, or Gaussian process regression (GPR)—may be employed. These models are capable of learning temporal dependencies and nonlinear transformations from raw biosignals to the corresponding muscle activation profiles and motion output parameters. Such models can further be trained to estimate motor intent, muscle force, physiological patterns and other neuromuscular parameters such as:
• Muscle force output,
• Joint torque,
• Kinematic trajectories, and
• Fine motor control indicators

by mapping the electrophysiological input features to biomechanical outputs. Training can be performed using supervised learning on annotated motion datasets or through semi-supervised/federated learning frameworks where data privacy and personalization are important. Collectively, these modeling approaches form the computational backbone of the system, enabling accurate interpretation of neuromuscular data and effective replication of expert-level motor performance in trainees through personalized feedback and adaptive training loops.

In one exemplary embodiment, the electrical activity of the trainer’s muscles, as recorded through EMG or sEMG sensors, is utilized as input for a neural network model designed to perform muscle activity pattern recognition and classification. This classified activity is then transferred to the trainee as part of a skill acquisition framework. The goal is to identify distinct neuromuscular signatures associated with expert-level movements and reproduce them in the trainee through controlled stimulation, guided motion, or haptic feedback. To this end, biosignals such as EEG, sEMG, or other electrophysiological signals, which jointly reflect the brain’s cognitive intent and the muscular execution patterns, are captured, stored, and subjected to a comprehensive feature extraction pipeline.

This processing is essential to isolate key time-domain, frequency-domain, and time-frequency domain characteristics that reflect variations in amplitude, firing rate, muscle activation onset, signal entropy, and spatial-temporal correlations. A critical observation in signal dynamics is that EEG signals exhibit significantly higher temporal variability and faster onset latency than EMG signals. EEG responses often precede EMG activity by several milliseconds to seconds, especially during volitional movements, due to the time it takes for descending motor commands to propagate from the motor cortex to peripheral muscle units.

As such, it becomes inherently challenging to correlate rapid EEG changes with downstream muscle activity using raw time-domain representations alone. Simple temporal alignment may fail to capture the causal relationship between intention and execution, especially in complex multi-joint tasks. Therefore, in accordance with one preferred embodiment, the system employs advanced time-frequency transformation techniques, such as:
• Short-time Fourier Transform (STFT),
• Continuous Wavelet Transform (CWT), or
• Hilbert-Huang Transform (HHT),
to decompose EEG signals into localized frequency bands that can be temporally aligned with corresponding EMG segments. This conversion allows for the identification of event-related desynchronization (ERD) or synchronization (ERS) patterns within specific brainwave bands (e.g., alpha, beta, gamma) that are indicative of motor preparation and execution.

By extracting synchronized or causally linked features from both EEG and sEMG domains—such as coherence, cross-correlation, mutual information—the system can model the neural-to-muscle transformation pathway with high fidelity. These multimodal features are then fed into deep learning models (e.g., hybrid CNN-LSTM architectures or transformer models with cross-attention) capable of classifying motor intents and predicting specific muscle activation patterns. This multimodal, temporally-aware processing not only improves the accuracy of motion intention prediction, but also enables phase-aligned skill transfer, wherein the timing and intensity of the trainee’s muscle stimulation or feedback are synchronized with the trainer’s original activation profile. Such a system architecture provides the foundational layer for personalized neuro-muscular training, where expert-level motion can be not just demonstrated but transferred at the level of neural and muscular activation patterns, ensuring accelerated learning, higher precision, and retention of fine motor skills in the trainee.

In other exemplary embodiment, a convolutional neural network (CNN) architecture is employed to process biosignals such as EEG, EMG, sEMG, or hybrid combinations thereof for the purpose of feature extraction and classification of muscle activity patterns. The CNN model is particularly well-suited for this application due to its hierarchical feature learning capability and spatial invariance, making it effective in capturing both localized and global patterns within raw or pre-processed bio-signal data.

Precisely, the CNN comprises of three essential layers: convolution layer, pooling layer, and fully connected layer.
a) Convolutional Layer – This layer applies a set of learnable filters (kernels) to the input biosignal. Each filter detects specific local patterns, such as frequency bursts, waveform transitions, or spike-timing features associated with neuromuscular events. In time-series biosignals, the convolution operation can highlight temporal motifs characteristic of distinct muscle contractions or neural firing sequences.
b) Pooling Layer – Typically a max-pooling or average-pooling layer, it reduces the dimensionality of the feature maps by summarizing adjacent feature responses. This enhances computational efficiency, introduces translation invariance, and mitigates the impact of small signal shifts or noise, which is particularly useful for real-world, non-stationary biosignals.
c) Fully Connected Layer (Dense Layer) – The flattened output of the final pooling layer is passed to fully connected layers, where high-level reasoning is performed. These layers combine and weight the extracted features to perform classification—i.e., associating the biosignal segment with a specific muscle activity label or motor class.

During training, the CNN learns to associate patterns in the input data with known motion labels by minimizing a loss function (e.g., categorical cross-entropy) via backpropagation and stochastic gradient descent (or its variants like Adam). Through iterative optimization, the network builds a model that can generalize across subjects or sessions, depending on training data diversity. In one exemplary embodiment, the classification of muscle activity is based on the morphological shape, frequency content, and temporal dynamics of the input biosignal. For example:
• In sEMG, different types of muscle contractions (e.g., isometric vs isotonic) produce distinct waveform shapes, RMS energy levels, and firing synchrony.
• In EEG, specific frequency bands (e.g., mu [8–13 Hz], beta [13–30 Hz]) undergo desynchronization (ERD) or synchronization (ERS) in response to motor planning or execution, which can serve as biomarkers for movement intention.

The process is grounded in the neurophysiological principle that voluntary movement originates in the motor cortex, where groups of neurons are activated in specific spatial-temporal patterns. The coordinated firing of these neurons generates fluctuating electrical potentials measurable as EEG signals.

Consequently, the variation in brain signal output during muscle activation can be mapped to distinct movement classes. For instance, EEG patterns associated with wrist flexion differ systematically from those of finger extension or grip adjustment. When these EEG or sEMG signals are processed through the CNN, the network classifies them based on learned templates of signal patterns corresponding to different muscle groups or gestures. This architecture enables the system to function as a neuro-muscular decoder, which not only classifies ongoing activity but can also be extended for gesture recognition, kinematic prediction, skill proficiency assessment, and neurofeedback-based rehabilitation or training loops.

Following the successful classification of the trainer’s muscle activity through bio-signal processing and deep learning models (e.g., CNN-based architecture), the system actuates the corresponding motor response in the trainee via neurostimulation techniques. This closes the sensorimotor loop, enabling real-time skill transfer by replicating both neural intent and musculoskeletal execution. In one working embodiment, Transcranial Magnetic Stimulation (TMS) is employed as a non-invasive neuromodulation technique to induce targeted excitability in the motor cortex of the trainee. TMS operates by delivering brief, focused magnetic pulses through a coil positioned over the scalp. These pulses generate induced electric currents in cortical neurons, especially in the primary motor cortex (M1), causing action potentials that propagate down the corticospinal tract to elicit voluntary-like muscle contractions.

The triggering of TMS pulses is dynamically controlled based on the classified biosignal of the trainer. The intensity, frequency, and localization of TMS can be calibrated to:
• Increase cortical excitability (high-frequency rTMS),
• Suppress competing pathways (low-frequency TMS),
• Induce long-term potentiation (LTP) or depression (LTD)-like effects for motor learning and plasticity, and
• Activate specific muscle representations mapped via motor evoked potentials (MEPs).

In an alternate embodiment, the system uses peripheral stimulation methods such as:
a) Transcutaneous Electrical Nerve Stimulation (TENS), or
b) Neuromuscular Electrical Stimulation (NMES),
to directly excite motor and sensory nerves at the level of the peripheral nervous system. These modalities are particularly effective for inducing controlled muscle contractions in the trainee’s limbs or target muscle groups, thereby mimicking the trainer’s actions at the neuromuscular level. The characteristics of the electrical stimulation—including pulse width, frequency, current amplitude, waveform shape, duty cycle, and stimulation site—are modulated in real-time to match the amplitude and timing profiles of the trainer’s muscle activity. This ensures that:
• Muscle synergies are accurately recreated,
• Joint trajectories are preserved,
• Proprioceptive feedback is naturally stimulated, and
• Muscle memory is progressively embedded in the trainee’s sensorimotor system.

Advanced implementations may employ closed-loop control, where real-time biofeedback (e.g., EMG from the trainee) is monitored to adjust stimulation parameters dynamically, ensuring synchronization, adaptation, and avoidance of fatigue or overstimulation. These neurostimulation approaches collectively enable the system to function as a neuro-motor transference interface, where encoded motor skills are effectively transferred not only through visual or verbal instruction but via direct cortical or peripheral activation, reinforcing motor learning through multisensory and neuromuscular channels.

In one exemplary scenario, following classification of the trainer’s motor intent and translation into a corresponding stimulation command, neuromuscular electrical stimulation (NMES) is applied to the trainee’s targeted muscle groups. This module operates under a controlled stimulation paradigm, wherein muscle contraction is evoked by activating efferent motor neurons through transcutaneous or intramuscular electric pulses, leading to movement generation and sensory feedback.

The strength and characteristics of the muscle contraction depend directly on the stimulation parameters, which can be dynamically modulated based on the complexity and intensity of the skill being transferred. Specifically:
a) Stimulation Intensity (Amplitude): Higher current amplitudes recruit more motor units by depolarizing a greater number of motor axons, leading to increased muscle fiber activation and stronger contraction force.
b) Pulse Duration (Width): Wider pulses allow more charge to be delivered, influencing the threshold of neuronal activation, especially for larger-diameter motor neurons that innervate fast-twitch muscle fibers.
c) Stimulation Frequency: Modulating the frequency (typically 20–100 Hz) determines the temporal summation of contractions, ranging from twitches to sustained tetanic contractions, enabling precise control over the speed and smoothness of the induced motion.
d) Waveform Shape: Biphasic or monophasic square pulses are commonly used; however, custom waveform shaping can enhance user comfort and optimize selective muscle targeting.

This computer-controlled electrical excitation of motor neurons mimics natural voluntary activation via descending cortical commands. The functional outcome—be it wrist extension, finger pinching, or joint rotation—is biomechanically indistinguishable from a consciously initiated movement, because the same peripheral motor pathway is engaged. Importantly, repeated and task-specific NMES leads to plastic changes in both spinal and supraspinal circuits. These activity-dependent plasticity effects contribute to the encoding of procedural memory, enabling the trainee to internalize motor patterns over time—even in the absence of stimulation. Thus, the NMES not only facilitates immediate motor replication but also acts as a neural trainer, reinforcing the motor engram through Hebbian principles ("neurons that fire together, wire together"). With each training iteration, the neural pathway between sensory input and motor output becomes more efficient, allowing the trainee to gradually reproduce the skill independently, without requiring external electrical cues.

The system further comprises a signal generation subsystem configured to emit a plurality of precisely calibrated electrical signals, wherein each signal is characterized by one or more parameters selected from location of application, temporal onset, pulse duration (width), frequency, and amplitude. These signal characteristics are intelligently modulated to induce either:
• A predetermined somatosensory cue (e.g., tactile feedback, vibration perception, proprioceptive stimulus), or
• A targeted motor event (e.g., joint flexion, fine motor grip, or limb extension), in the user (trainee).

The electrical signal receptors involved in this stimulation-induced response include cutaneous mechanoreceptors, proprioceptors within muscle spindles, Golgi tendon organs, peripheral motor neurons, and sensory afferents. These receptors transduce the externally applied electrical signals into physiological responses by activating underlying neural and muscular circuits.

The signal generator may optionally be coupled with multi-modal feedback modules, capable of delivering first-level sensory cues (e.g., vibrotactile, auditory, or visual) that serve as preparatory or associative reinforcement signals during training. These cues can help encode associative learning patterns. The intensity of the neuromuscular response—including the degree of contraction, perceived force, or joint angular displacement—is proportional to the amplitude and spatial focus of the applied electrical signal. In high-precision applications, multi-electrode stimulation arrays are used to deliver spatially selective and graded stimulation to deeper muscle groups.

Crucially, the stimulation triggers automatic and involuntary contraction of the target muscle, closely mimicking the natural reflex arc or efferent-voluntary action. This “auto-action response” is elicited through targeted activation of motor endplates via stimulation of peripheral nerve endings, thereby bypassing higher cortical initiation. As a result, the movement executed is biomechanically identical to voluntary motion, yet driven externally by the skill transfer system. Furthermore, each signal’s characteristic—location (electrode placement), timing (inter-stimulus interval), pulse length (e.g., 50–400 µs), frequency (e.g., 20–100 Hz), and amplitude (e.g., 5–120 mA)—is dynamically adjusted by the system controller in real time. This modulation depends on:

• The type of muscle fiber being targeted (slow vs. fast twitch),
• The desired functional movement outcome,
• The user’s physiological adaptation profile (muscle fatigue, neural recruitment threshold), and
• The training phase (initiation, repetition, consolidation).

The closed-loop control logic, optionally supported by feedback from EMG sensors or inertial measurement units (IMUs) placed on the trainee, ensures that the output movement correlates with intended skill pattern, both in force profile and temporal fidelity. This adaptive, signal-driven, and feedback-coupled stimulation architecture allows the system to serve as a robust neuro-muscular proxy for skill encoding, whereby the sensory-motor experience of the trainer is precisely reconstructed in the trainee, not only facilitating motor learning, but also enabling procedural memory consolidation via repetitive exposure to accurate neuromuscular cues.

Following the initial stimulation and response phase, the system enters a validation and feedback loop, wherein the muscle activity induced at the trainee’s end is recorded using the same bioelectrical signal acquisition setup as deployed at the trainer’s end. This mirror-mode acquisition is crucial to determine whether the simulated motor action at the trainee side replicates the trainer’s original movement profile in terms of:
• Activation intensity (measured via EMG signal amplitude),
• Contraction force (using force sensors or torque encoders),
• Range of motion or angular displacement (captured via inertial measurement units or motion capture), and
• Temporal dynamics (latency, duration, rise/fall time of muscle activation)

In accordance with one exemplary embodiment, transcutaneous electrical signals such as TENS-type (Transcutaneous Electrical Nerve Stimulation) or NMES-type (Neuromuscular Electrical Stimulation) pulses are applied to the trainee’s muscle group by generating a controlled DC voltage across electrode groups. Further, the system supports a calibration mode, wherein: the electrode groups are initially used on the trainer, not just to stimulate but also to record bioelectrical signals during execution of known reference movements. These reference signals include both myoelectric patterns (sEMG) and neural response features, and are tagged with corresponding movement annotations (e.g., flexion, grip, extension).

During calibration at the trainee’s end, the same muscle group and anatomical coordinates are targeted with incremental adjustments in:
• Electrode positioning,
• Signal amplitude and frequency, and
• Stimulation waveform (monophasic, biphasic, triangular, exponential decay).

The goal is to evoke a muscle response in the trainee that matches the recorded trainer reference movement as closely as possible. This process is governed by closed-loop control algorithms that monitor real-time EMG signals and joint kinematics from the trainee and compare them to the trainer's signature movement features. Additionally, the system may support adaptive learning protocols, wherein the electrode stimulation parameters are fine-tuned dynamically via reinforcement signals derived from movement accuracy metrics, and machine learning models—such as recurrent neural networks (RNNs) or autoencoders—are employed to map between the trainer's activation domain and the trainee's muscular topology, compensating for inter-individual variability.

Moreover, in one alternate embodiment, a lookup table or parametric response model is constructed during initial calibration, capturing the relationship between specific electrode configurations and resultant muscle outcomes. This model enables rapid and precise invocation of desired responses in the trainee upon receiving new input signals from the trainer or from pre-recorded expert datasets.

The ability to record, compare, and iteratively calibrate both the input (signal) and output (motor effect) in this bi-directional manner ensures that the transferred skill is faithful to the trainer’s biomechanical expression, and the neuroplastic reinforcement loop at the trainee’s end is optimized for motor learning, adaptation, and eventual autonomous skill execution.

In one example embodiment, deep learning algorithms are employed to dynamically modulate the parameters of neuromuscular electrical stimulation (NMES), enabling precise induction of muscle contractions that replicate the motor actions originally executed by a trainer. These algorithms may utilize sensor feedback and biomechanical data to tailor stimulation profiles—such as pulse amplitude, width, frequency, and waveform symmetry—thereby eliciting targeted and repeatable movements in specific muscle groups of the trainee, such as those in the arm or fingers. The resultant neuromuscular responses simulate voluntary movement with high fidelity, despite being involuntarily induced.

To facilitate performance evaluation and real-time feedback, the system may incorporate embedded transducers or inertial measurement units (IMUs) within a wearable interface (e.g., a sleeve). These sensors capture spatiotemporal metrics, such as joint angles, displacement vectors, muscle deformation patterns, and force dynamics. The acquired kinematic and kinetic data can be analysed to assess alignment between trainer-induced and trainee-executed motions.

The NMES employed typically delivers low-frequency (e.g., 20–50 Hz), high-intensity biphasic pulses, designed to activate alpha motor neurons while minimizing tissue fatigue and charge accumulation. Biphasic stimulation—comprising balanced cathodic and anodic phases—helps prevent electrochemical damage at the electrode-skin interface and supports sustained, functional contractions. By selectively recruiting motor units in a manner guided by machine learning models trained on expert motion datasets, the system can achieve a high degree of motor mimicry and neuromuscular re-education, useful in physical rehabilitation, skill training, and teleoperated kinesthetic transfer systems.

In a more specific embodiment, analytics pertaining to muscle memory acquisition and transfer are continuously computed and visualized via a head-mounted display (HMD) system worn by both the trainer and the trainee. This bidirectional visual feedback interface allows real-time monitoring of neuromotor synchronization and training progress. The system computes a muscle memory congruence score by comparing the spatiotemporal motion signatures and electromyographic (EMG) response profiles of the trainee against those of the trainer, using cross-correlation, dynamic time warping, or other similarity metrics.

The congruence score, representing the degree of fidelity in the neuromotor replication, is benchmarked against a predetermined threshold. Should the score fall below this threshold, indicating suboptimal motor learning or execution variance, the system autonomously adapts the neuromuscular electrical stimulation (NMES) parameters at the trainee’s end. This adaptive modulation may involve fine-tuning stimulation intensity (mA), pulse duration (µs), frequency (Hz), or recruitment pattern to reinforce correct proprioceptive feedback loops and facilitate more accurate muscle fiber activation.

Additionally, the system may incorporate reinforcement learning algorithms to iteratively refine stimulation protocols based on individual trainee responsiveness and progress trajectories. Over successive training iterations, the HMD may also overlay augmented visual cues or haptic feedback to enhance motor learning via multimodal sensory integration. This closed-loop system thus not only ensures high-resolution motor memory transfer but also supports autonomous calibration of stimulation profiles to optimize neuromuscular re-education and performance convergence.

Next, a method 1000 for skill transfer via a human-to-human neurophysiological interface is disclosed. This method enables the replication of fine motor skills from a trainer to a trainee by capturing and translating electrophysiological biosignals into targeted electrical stimulation commands. Step 110 involves the acquisition of biosignals such as electroencephalography (EEG), electromyography (EMG) and surface electromyography (sEMG) from a plurality of trainers. These biosignals are recorded in real time while the trainers perform specific motor tasks at an individual level. The signals reflect both cortical motor intent (EEG) and peripheral muscle activity (EMG), thereby capturing a comprehensive neural-muscular signature of each action.

In Step 210, the captured EEG/EMG datasets undergo pre-processing to enhance signal quality. This involves removal of artifacts (e.g., motion, ocular, and environmental noise) using one or more machine learning models. The cleaned data is then segmented and normalized to facilitate uniform feature extraction. Step 310 applies machine learning (ML) or deep learning (DL) algorithms to classify muscle activity based on temporal, spectral, and spatial patterns inherent in the EEG/EMG signals. Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), or hybrid architectures like CNN-LSTM models may be utilized to detect and differentiate between subtle variations in muscle activation patterns. The trained models classify signal segments corresponding to discrete motor actions and quantify their intensity, directionality, and duration.

In Step 410, the output of the classification model is used to generate neuromodulatory stimulation protocols tailored for the trainee. Specifically, Transcutaneous Electrical Nerve Stimulation (TENS) or Neuromuscular Electrical Stimulation (NMES) is applied to activate the trainee’s corresponding nerve pathways and muscle groups. These stimulation protocols mirror the classified motor patterns from the trainer, effectively inducing analogous muscle contractions and proprioceptive feedback at the trainee’s end.

Step 510 introduces a performance evaluation loop. The motor output generated at the trainee’s end—whether tracked via motion capture systems, inertial sensors, or secondary EMG/EEG recordings—is compared against the trainer's original movement. This comparison may utilize metrics such as joint trajectory similarity, movement velocity, and activation timing, resulting in the computation of a performance score. This score quantitatively reflects the accuracy, latency, and coordination of the trainee’s motor reproduction.

Step 610 involves an adaptive recalibration of the stimulation parameters based on the performance score. If the trainee exhibits underperformance or incomplete motor actuation, the system dynamically adjusts the stimulation magnitude, pulse width, or frequency to better elicit the desired response. For example, if the task demands a high-speed or high-precision action, higher amplitude NMES pulses may be applied to ensure robust motor unit recruitment and to facilitate muscle memory consolidation.

Over iterative cycles, this closed-loop feedback system enables personalized motor training, neuroplastic adaptation, and long-term skill acquisition through neural entrainment. The disclosed method is particularly suited for applications in neurorehabilitation, prosthetics training, sports coaching, and remote motor skill transfer in high-precision professions.
In accordance with an embodiment, the head mounted display 220, 250 comprises a memory unit configured to store machine-readable instructions. The machine-readable instructions may be loaded into the memory unit from a non-transitory machine-readable medium, such as, but not limited to, CD-ROMs, DVD-ROMs and Flash Drives. Alternately, the machine-readable instructions may be loaded in a form of a computer software program into the memory unit. The memory unit in that manner may be selected from a group comprising EPROM, EEPROM and Flash memory. Further, the camera processing module includes a processor operably connected with the memory unit. In various embodiments, the processor is one of, but not limited to, a general-purpose processor, an application specific integrated circuit (ASIC) and a field-programmable gate array (FPGA).

In general, the word “module,” as used herein, refers to logic embodied in hardware or firmware, or to a collection of software instructions, written in a programming language, such as, for example, Java, C, or assembly. One or more software instructions in the modules may be embedded in firmware, such as an EPROM. It will be appreciated that modules may comprised connected logic units, such as gates and flip-flops, and may comprise programmable units, such as programmable gate arrays or processors. The modules described herein may be implemented as either software and/or hardware modules and may be stored in any type of computer-readable medium or other computer storage device.

Further, while one or more operations have been described as being performed by or otherwise related to certain modules, devices or entities, the operations may be performed by or otherwise related to any module, device or entity. As such, any function or operation that has been described as being performed by a module could alternatively be performed by a different server, by the cloud computing platform, or a combination thereof. It should be understood that the techniques of the present disclosure might be implemented using a variety of technologies. For example, the methods described herein may be implemented by a series of computer executable instructions residing on a suitable computer readable medium. Suitable computer readable media may include volatile (e.g., RAM) and/or non-volatile (e.g., ROM, disk) memory, carrier waves and transmission media. Exemplary carrier waves may take the form of electrical, electromagnetic or optical signals conveying digital data steams along a local network or a publicly accessible network such as the Internet.

It should also be understood that, unless specifically stated otherwise as apparent from the following discussion, it is appreciated that throughout the description, discussions utilizing terms such as "controlling" or "obtaining" or "computing" or "storing" or "receiving" or "determining" or the like, refer to the action and processes of a computer system, or similar electronic computing device, that processes and transforms data represented as physical (electronic) quantities within the computer system's registers and memories into other data similarly represented as physical quantities within the computer system memories or registers or other such information storage, transmission or display devices. Various modifications to these embodiments are apparent to those skilled in the art from the description and the accompanying drawings. The principles associated with the various embodiments described herein may be applied to other embodiments. Therefore, the description is not intended to be limited to the embodiments shown along with the accompanying drawings but is to be providing broadest scope of consistent with the principles and the novel and inventive features disclosed or suggested herein. Accordingly, the invention is anticipated to hold on to all other such alternatives, modifications, and variations that fall within the scope of the present invention.
,CLAIMS:We Claim:

1) A method for enabling real-time skill transfer between a trainer and trainee, the method comprising:
acquiring electrophysiological signals from one or more trainer(s) via a set of non-invasive, high-sensitivity electrodes embedded within a head mounted device;
pre-processing the acquired electrophysiological signals and analysing the pre-processed electrophysiological signals using one or more machine learning models to estimate motor intent, muscle force, identify physiological patterns and extract one or more neuromuscular parameters;
classifying the analysed electrophysiological signals into a corresponding muscle activity label; and
triggering a neurostimulation protocol in the trainee based on the classified muscle activity, wherein the stimulation is configured to replicate trainer’s muscle activation pattern in the trainee.

2) The method, as claimed in accordance with claim 1, wherein the electrophysiological signals comprise of surface electroencephalography (EEG), electromyography (EMG), surface electromyography (sEMG), and sensory evoked potentials (SEPs).

3) The method, as claimed in accordance with claim 2, wherein the electrophysiological signals are captured from multiple neuroanatomical loci comprising a cerebral cortex, brain stem, spinal cord, peripheral nerves and a combination thereof.

4) The method, as claimed in accordance with claim 1, wherein the acquired electrophysiological signals are pre-processed by applying band-pass and/or notch filters to isolate relevant signal frequencies, performing full-wave rectification of the electrophysiological signals, conducting wavelet transform-based time-frequency analysis, applying smoothing techniques to reduce variability, and optionally extracting spectral features using Fast Fourier Transform (FFT).

5) The method, as claimed in accordance with claim 1, wherein the machine learning models are selected from a combination of convolutional neural networks (CNNs) for spatial feature extraction, and recurrent neural networks (RNNs), optionally long short-term memory (LSTM) networks, for temporal sequence analysis of muscle activation patterns from the electrophysiological signals.

6) The method, as claimed in accordance with claim 5, wherein the convolutional neural networks (CNNs) is utilized for the classification based on morphological, spectral and temporal characteristics of the electrophysiological signals.

7) The method, as claimed in accordance with claim 1, wherein the neurostimulation by way of electrically evoked contraction is triggered via at least one of Transcranial Magnetic Stimulation (TMS) or Neuromuscular Electrical Stimulation (NMES) to directly excite target motor nerves or muscle groups in the trainee corresponding to the trainer’s muscle activation patterns.

8) The method, as claimed in accordance with claim 7, wherein one or more neurostimulation parameters such as pulse width, frequency, current amplitude, waveform shape, duty cycle, and stimulation site are modulated in real time to match amplitude and timing profile of the trainer’s muscle activation patterns.

9) The method, as claimed in accordance with claim 2, further comprising decomposing the EEG signals into localized frequency bands such that the EEG signals are temporally aligned with corresponding EMG signals or sEMG signals and the neural-to-muscle transformation is achieved in high fidelity, wherein the EEG signals are decomposed using techniques such as Short-time Fourier Transform (STFT), Continuous Wavelet Transform (CWT) or Hilbert-Huang Transform (HHT).

10) The method, as claimed in accordance with claim 7, further comprising generating movement and proprioceptive sensory feedback in the trainee as a result of the electrically evoked contraction triggered using the neurostimulation.

11) The method, as claimed in accordance with claim 8, wherein the one or more neurostimulation parameters are adjusted adaptively based on trainee’s muscle performance and signal fidelity.

12) A system (1000) for enabling real-time skill transfer between a trainer and trainee, the system (1000) comprising:
a head mounted device (200) worn by the trainer and the trainee, and configured to acquire electrophysiological signals from one or more trainer(s) via a set of non-invasive, high-sensitivity electrodes (300) embedded within the head mounted device (200);
a processing module (500) configured to:
pre-process the acquired electrophysiological signals and analyse the pre-processed electrophysiological signals using one or more machine learning models to estimate motor intent, muscle force, identify physiological patterns and extract one or more neuromuscular parameters;
classify the analysed electrophysiological signals into a corresponding muscle activity label; and
trigger a neurostimulation protocol in the trainee based on the classified muscle activity, wherein the stimulation is configured to replicate trainer’s muscle activation pattern in the trainee.

13) The system (1000), as claimed in accordance with claim 12, wherein the electrophysiological signals comprise of surface electroencephalography (EEG), electromyography (EMG), surface electromyography (sEMG), and sensory evoked potentials (SEPs).

14) The system (1000), as claimed in accordance with claim 2, wherein the electrophysiological signals are captured from multiple neuroanatomical loci comprising a cerebral cortex, brain stem, spinal cord, peripheral nerves and a combination thereof.

15) The system (1000), as claimed in accordance with claim 12, wherein the processing module (500) is configured to pre-process the acquired electrophysiological signals by applying band-pass and/or notch filters to isolate relevant signal frequencies, performing full-wave rectification of the electrophysiological signals, conducting wavelet transform-based time-frequency analysis, applying smoothing techniques to reduce variability, and optionally extracting spectral features using Fast Fourier Transform (FFT).

16) The system (1000), as claimed in accordance with claim 12, wherein the machine learning models are selected from a combination of convolutional neural networks (CNNs) for spatial feature extraction, and recurrent neural networks (RNNs), optionally long short-term memory (LSTM) networks, for temporal sequence analysis of muscle activation patterns from the electrophysiological signals.

17) The system (1000), as claimed in accordance with claim 16, wherein the convolutional neural networks (CNNs) is utilized for the classification based on morphological, spectral and temporal characteristics of the electrophysiological signals.

18) The system (1000), as claimed in accordance with claim 12, wherein the neurostimulation by way of electrically evoked contraction is triggered via at least one of Transcranial Magnetic Stimulation (TMS) or Neuromuscular Electrical Stimulation (NMES) to directly excite target motor nerves or muscle groups in the trainee corresponding to the trainer’s muscle activation patterns.

19) The system (1000), as claimed in accordance with claim 18, wherein one or more neurostimulation parameters such as pulse width, frequency, current amplitude, waveform shape, duty cycle, and stimulation site are modulated in real time to match amplitude and timing profile of the trainer’s muscle activation patterns.

20) The system (1000), as claimed in accordance with claim 13, further comprising decomposing the EEG signals into localized frequency bands such that the EEG signals are temporally aligned with corresponding EMG signals or sEMG signals and the neural-to-muscle transformation is achieved in high fidelity, wherein the EEG signals are decomposed using techniques such as Short-time Fourier Transform (STFT), Continuous Wavelet Transform (CWT) or Hilbert-Huang Transform (HHT).

21) The system (1000), as claimed in accordance with claim 18, further comprising generating movement and proprioceptive sensory feedback in the trainee as a result of the electrically evoked contraction triggered using the neurostimulation.

22) The system (1000), as claimed in accordance with claim 19, wherein the one or more neurostimulation parameters are adjusted adaptively based on trainee’s muscle performance and signal fidelity.