A physiological-based robot speed control method, device, equipment and medium

Abstract

The application belongs to the technical field of data processing, and discloses a robot speed control method and device based on physiology, equipment and medium, the method comprises the following steps: acquiring an original physiological signal and a video stimulation intensity curve; establishing an individual physiological response model and a standardized mapping function according to the original physiological signal and the video stimulation intensity curve; acquiring real-time multi-modal signals, inputting the real-time multi-modal signals into the individual physiological response model and the standardized mapping function and calculating to obtain a risk score; calculating according to the risk score to obtain a speed modulation coefficient; and performing robot speed control operation according to the speed modulation coefficient. The application can improve the comfort degree of the cooperation process and the overall execution efficiency of the robot operation on the basis of ensuring the safety of human-machine cooperation.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a physiological-based method, device, equipment, and medium for robot speed control. Background Technology

[0002] With the continuous development of industrial robot technology, human-robot collaborative operation is being applied more and more widely in industrial production scenarios. Robot safety control technology is the core foundation for ensuring the stable and reliable operation of human-robot collaboration. Currently, traditional robot safety control methods in the industrial field mainly focus on physical safety protection. Among them, the Speed ​​and Separation Monitoring (SSM) mechanism specified in ISO 15066 is the most widely used. This method collects human-robot spatial distance information in real time, adjusts the robot's operating speed according to a preset distance-speed mapping rule, gradually reduces the robot speed when the human-robot distance shrinks to a safe threshold, and triggers shutdown protection when the distance reaches a dangerous threshold, thereby avoiding physical collision accidents between humans and robots. This is the mainstream technical solution for human-robot collaborative safety management at present.

[0003] However, existing physical safety control methods have significant limitations. First, these methods generally employ fixed speed thresholds and mapping rules, failing to adapt to individual differences among operators. This can easily induce panic in operators with high psychological sensitivity, while imposing overly conservative control restrictions on operators with lower psychological sensitivity, neglecting the operator's psychological safety needs throughout the process. Second, the simple discrete distance-velocity mapping mechanism is prone to frequent switching of the robot's operating state at threshold boundaries, causing robot motion jitter, which reduces the efficiency of task execution and affects the operator's experience. Third, existing control modes are passive reactive controls, only implementing deceleration actions after a dangerous situation arises where the human-machine distance approaches, failing to anticipate and respond to the operator's accumulated stress. In closely collaborative work scenarios, physiological signals such as the operator's pupil diameter can dynamically reflect their level of tension, but existing safety control systems have not yet established a mechanism to effectively translate these physiological signals into robot behavior adjustment strategies. This can easily lead to operators being in a state of high concentration for extended periods, adversely affecting their work efficiency and physical and mental health.

[0004] Therefore, ensuring the physical safety of human-machine collaboration and improving its efficiency and operational comfort have become urgent problems to be solved. Summary of the Invention

[0005] This application provides a physiologically based robot speed control method, device, equipment, and medium, which can complete individual-based calibration through raw physiological signals and video stimulus intensity curves. Furthermore, it establishes a physiological response model and standardized mapping function to adapt to the differences in physiological responses among different operators, eliminating judgment bias caused by individual baseline differences. Real-time acquisition of multimodal signals input to the model and mapping function, and calculation of risk scores, allows for the simultaneous fusion of multi-dimensional information, ensuring that risk assessment aligns with the operator's actual situation. Continuous speed modulation coefficients are output based on the risk score, enabling smooth transitions in robot speed adjustments and avoiding motion fluctuations caused by frequent state switching. This application can improve the comfort of the collaborative process while ensuring human-robot collaboration safety, and enhance the overall execution efficiency of robot operations.

[0006] In a first aspect, embodiments of this application provide a physiological-based robot speed control method, the method comprising: Acquire raw physiological signals and video stimulus intensity curves; Establish individual physiological response models and standardized mapping functions based on raw physiological signals and video stimulus intensity curves; Real-time multimodal signals are acquired, input into an individual physiological response model and a standardized mapping function, and calculated to obtain a risk score. The speed modulation coefficient is calculated based on the risk score; The robot speed is controlled based on the speed modulation coefficient.

[0007] Furthermore, based on the original physiological signals and video stimulus intensity curves, individual physiological response models and standardized mapping functions are established, including: A standardized mapping function was calculated based on the original physiological signals, the relaxation baseline value, and the upper limit of tension. An individual physiological response model is established based on calculations using the video stimulus intensity curve, gain coefficient, relaxation baseline value, and physiological response delay.

[0008] Furthermore, real-time multimodal signals are acquired, input into an individual physiological response model and a standardized mapping function, and calculated to obtain a risk score, including: Acquire real-time multimodal signals; real-time multimodal signals include standardized physiological signals, spatial signals, and signal quality coefficients; Real-time multimodal signals are input into an individual physiological response model and a standardized mapping function to perform time delay estimation and nonlinear least squares fitting, thereby obtaining the model predictions. The risk score is calculated based on the mean of the measured data and the model prediction.

[0009] Furthermore, based on the risk score, the speed modulation coefficient is calculated, including: Establish a mapping function from the risk score to the target speed coefficient based on the risk score; The hysteresis velocity coefficient is obtained based on the hysteresis bandwidth and the mapping function; The constraint velocity coefficient is calculated based on the actual output velocity and hysteresis velocity coefficient of the previous moment. The output speed coefficient is calculated based on the constraint speed coefficient. The output speed coefficient is processed based on the indicator function to obtain the speed modulation coefficient.

[0010] Furthermore, the raw physiological signal is the pupil diameter detected when the operator watches a standardized stimulus video of a preset duration.

[0011] Furthermore, the method also includes: Normalized visual motion velocity was calculated based on optical flow. The normalized acceleration is obtained based on the velocity difference between adjacent frames; The video stimulus intensity curve is constructed based on the weighting coefficients, normalized height, normalized visual motion velocity, and normalized acceleration.

[0012] Furthermore, the speed modulation coefficient ranges between 0 and 1; When the speed modulation coefficient is 1.0, the robot runs at full speed; The robot stops when the speed modulation coefficient is 0.0.

[0013] Secondly, embodiments of this application provide a physiological-based robot speed control device, the device comprising: The acquisition module is used to acquire raw physiological signals and video stimulus intensity curves; The module is used to build individual physiological response models and standardized mapping functions based on raw physiological signals and video stimulus intensity curves; The calculation module is used to acquire real-time multimodal signals, input the real-time multimodal signals into an individual physiological response model and a standardized mapping function, and calculate to obtain a risk score; The modulation module is used to calculate the speed modulation coefficient based on the risk score; The control module is used to control the robot's speed based on the speed modulation coefficient.

[0014] Thirdly, embodiments of this application provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it performs the steps of a physiological-based robot speed control method as described in any of the above embodiments.

[0015] Fourthly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a physiological-based robot speed control method as described in any of the above embodiments.

[0016] In summary, compared with the prior art, the beneficial effects of the technical solution provided in this application include at least the following: This application provides a physiologically-based robot speed control method that can perform individual-based calibration using raw physiological signals and video stimulus intensity curves. Furthermore, it establishes a physiological response model and a standardized mapping function to adapt to the differences in physiological responses among different operators, eliminating judgment biases caused by individual baseline differences. Real-time acquisition of multimodal signals input to the model and mapping function, and calculation of a risk score, allows for the simultaneous fusion of multi-dimensional information, ensuring that risk assessment aligns with the operator's actual situation. Continuous speed modulation coefficients are output based on the risk score, enabling smooth transitions in robot speed adjustments and avoiding motion fluctuations caused by frequent state switching. This application improves the comfort of the collaborative process while ensuring human-robot collaboration safety, and enhances the overall execution efficiency of robot operations. Attached Figure Description

[0017] Figure 1 A flowchart illustrating a physiological-based robot speed control method provided as an exemplary embodiment of this application.

[0018] Figure 2 This is a structural diagram of a physiologically based robot speed control device provided as an exemplary embodiment of this application. Detailed Implementation

[0019] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0020] Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] Please see Figure 1 This application provides a physiological-based robot speed control method, which specifically includes the following steps: Step S1: Obtain the original physiological signals and video stimulus intensity curves.

[0022] In some embodiments, the raw physiological signal is the pupil diameter detected when the operator watches a standardized stimulus video of a preset duration.

[0023] In one feasible implementation, the preset duration is 60 seconds.

[0024] The method employs a standardized stimulation video with a preset duration of 60 seconds as the stimulus carrier for data acquisition. This allows for the complete acquisition of pupil diameter data within a limited timeframe, avoiding operator fatigue or decreased attention due to excessively long acquisition times while ensuring the integrity of the stimulation process and covering all stages of stimulus change. Simultaneous acquisition of the raw physiological signal and the video stimulus intensity curve directly establishes a correspondence between physiological response data and external stimulus intensity, providing an accurate stimulus reference for subsequent analysis of physiological response patterns. Using pupil diameter as the raw physiological signal eliminates the need for invasive procedures, allowing operators to complete the test under natural viewing conditions, resulting in data that more closely reflects real physiological states. The 60-second duration adapts to general visual testing paradigms, facilitating standardized reuse across different scenarios and reducing data deviations caused by differences in test parameters.

[0025] In one feasible implementation, the operator's physiological signals in human-machine collaboration are influenced by both individual differences and environmental factors, making the raw measurements unsuitable for direct cross-person comparisons. Given an operator... primitive physiological signals (e.g., pupil diameter, unit: mm), the absolute value difference between individuals under the same stress state can reach over 40%. The standardized stress index is defined as: ; in, For operators The relaxation baseline value, This sets its upper limit for tension. The core issue then becomes: how to quickly and accurately identify each operator's tension through controlled stimuli. Parameter pair.

[0026] In some embodiments, the method further includes: Normalized visual motion velocity was calculated based on optical flow. The normalized acceleration is obtained based on the velocity difference between adjacent frames; The video stimulus intensity curve is constructed based on the weighting coefficients, normalized height, normalized visual motion velocity, and normalized acceleration.

[0027] In one feasible implementation, a 60-second segmented visual stimulus video (such as a first-person perspective of a roller coaster) is designed, with a stimulus intensity function... Defined as: ; in, The normalized visual motion velocity was calculated using the optical flow method. Normalized acceleration is obtained by the velocity difference between adjacent frames; Normalized altitude, representing the viewpoint's height above the ground; weighting coefficients satisfy It was calibrated based on ergonomic experimental data. A typical stimulus sequence consists of six phases: an initial static phase... (Establish relaxation baseline in 0-10s), Climbing phase (10-20 seconds of mild stimulation), dive phase (20-30 seconds of moderate stimulation), sharp turn segment (30-40 seconds of intense stimulation), weightlessness segment (Maximum stimulation in 40-50 seconds), deceleration phase (Restore baseline in 50-60 seconds).

[0028] Step S2: Establish an individual physiological response model and a standardized mapping function based on the original physiological signals and video stimulus intensity curves.

[0029] In some embodiments, an individual physiological response model and a standardized mapping function are established based on the original physiological signals and video stimulus intensity curves, including: A standardized mapping function was calculated based on the original physiological signals, the relaxation baseline value, and the upper limit of tension. An individual physiological response model is established based on calculations using the video stimulus intensity curve, gain coefficient, relaxation baseline value, and physiological response delay.

[0030] Specifically, a standardized mapping function is derived by combining raw physiological signals with relaxation baseline values ​​and tension upper limits. This smooths out inherent differences in the basic physiological states of different individuals, establishing a unified quantitative scale for physiological response data and avoiding interference from individual differences in baseline values ​​in subsequent analysis. An individual physiological response model is constructed by superimposing gain coefficients, relaxation baseline values, and physiological response delays onto the video stimulus intensity curve. This model can reconstruct the complete transmission process of stimulus acting on the physiological system and match the unique response change patterns of each individual. The two methods work together to achieve accurate two-way conversion between stimulus intensity and physiological response data, providing stable computational support for subsequent physiological state assessment. This balances individual differences with a unified evaluation scale, ensuring that the analysis results meet practical application needs.

[0031] In one feasible implementation, since the physiological signal response to visual stimuli has nonlinear and time-delay characteristics, a parameterized response model is established: ; in, for Pupil diameter at any given moment The baseline pupil value in an individual's relaxed state characterizes the individual's physiological characteristics; This is the gain coefficient, characterizing an individual's sensitivity to stimuli (highly sensitive operators). Low-sensitivity operators ), The physiological response delay reflects the time of nerve conduction and muscle response (typically ranging from 0.5 to 1.5 seconds). The nonlinear exponent characterizes the nonlinear relationship between stimulus intensity and response amplitude (typically ranging from 1.2 to 1.8). This model is constructed based on the nonlinear perception principle of the Weber-Fechner law.

[0032] In one feasible implementation, parameter identification employs a two-step optimization strategy: Step 1: Time Delay Estimation. The optimal delay is determined through cross-correlation analysis. : ; in, The Pearson correlation coefficient is defined as follows: Search scope The normal physiological delay range is covered by seconds, and the peak correlation coefficient is searched using the sliding window method with a step size of 0.1 seconds.

[0033] Step 2: Nonlinear least squares fitting. (Fixed) Solve later: ; in, The total number of sampling points corresponds to a 60-second sampling rate of 10Hz. For the first At each sampling time, the objective function To predict the sum of squared residuals.

[0034] In one feasible implementation, a real-time normalization formula is derived based on the identified response function. This is defined as the full stimulus response (…). The pupil diameter is: ; The real-time standardized value is: ; in, The original pupil diameter as measured currently. This is a standardized tension index, where 0 corresponds to complete relaxation and 1 corresponds to maximum tension. The physical meaning of this formula is: mapping an individual's current state to the relative position of their personal response range.

[0035] Furthermore, the quality of the fit is determined by the coefficient of determination. Evaluate: ; in, These are the model's predicted values. This is the average of the measured data. For the sum of squared residuals, The sum of squares is the total square. The quality judgment criterion is: if... If the data quality is deemed unacceptable, it needs to be re-collected. Possible reasons include operator negligence or sensor malfunction. Abnormal parameter detection includes: If... Marked as "extremely low sensitivity," manual confirmation is required for operator intervention; if If 's' is marked as "delay anomaly," the sensor time synchronization needs to be checked; if mm or mm is marked as "baseline abnormality", and interference factors such as wearing glasses need to be excluded.

[0036] Step S3: Acquire real-time multimodal signals, input the real-time multimodal signals into the individual physiological response model and standardized mapping function, and calculate to obtain a risk score.

[0037] In some embodiments, real-time multimodal signals are acquired, input into an individual physiological response model and a standardized mapping function, and calculated to obtain a risk score, including: Acquire real-time multimodal signals; real-time multimodal signals include standardized physiological signals, spatial signals, and signal quality coefficients; Real-time multimodal signals are input into an individual physiological response model and a standardized mapping function to perform time delay estimation and nonlinear least squares fitting, thereby obtaining the model predictions. The risk score is calculated based on the mean of the measured data and the model prediction.

[0038] The process involves collecting real-time multimodal signals, including standardized physiological signals, spatial signals, and signal quality coefficients. This allows for the acquisition of real-time status information from multiple dimensions, including physiological performance, spatial state, and signal quality, providing multidimensional data support for risk assessment. After inputting the real-time multimodal signals into an individual physiological response model and a standardized mapping function, model predictions are obtained through time delay estimation and nonlinear least squares fitting. This ensures that the model output closely reflects the rhythm and patterns of real-time physiological changes, reducing the deviation between calculated results and actual conditions. Combining the mean of measured data with model predictions yields a risk score, taking into account both individual response patterns and real-time measured data, making the risk assessment results more closely reflect the true state of the tested individual.

[0039] In one feasible implementation, a signal quality coefficient is defined. , characterizing the Reliability of each signal source: ; in, The variance of the signal within a 1-second sliding window characterizes the signal stability. The difference between adjacent frames characterizes the degree of abrupt change. and For the quality threshold (for pupil signals), mm², (mm). The signal quality coefficient can identify abnormal situations such as sensor tracking failure and blink interference.

[0040] Furthermore, define the integration risk score. : ; Linear weighted part: ; in, The total number of signal sources, For the first Preset weights for each signal (recommended configuration: physiological signals) Distance signal auxiliary signal ), For signal quality coefficients, This is the standardized signal value. Denominator normalization ensures the total weight is 1.

[0041] Specifically, linear weighting is used to fuse multi-source risk signals. Each weight represents the relative contribution of different signals to psychological safety risk, and denominator normalization ensures that the results are scale-stable, interpretable, and easy to calculate in real time.

[0042] Auxiliary signals may include robot speed / acceleration, end-effector direction, task stage, task mode, environmental congestion, historical alarms, or human intervention records. These auxiliary signals are considered to supplement physiological signals and human-robot distance, which cannot fully reflect task and environmental risks, thus reducing misjudgments.

[0043] Interactive items: ; in, Physiological tension, Spatial proximity (defined as) , Human-machine distance, cm, cm), interaction item weight Calibrated based on experimental data. The physical meaning of the interaction term is: when the operator is highly stressed and very close to the robot, the risk does not simply add up linearly but increases exponentially; this term captures the compound threat effect.

[0044] Among them, the interaction term uses a standardized physiological tension index and a spatial proximity index multiplied together. The reason is that when the operator is tense and the distance between the human and the machine is close, the psychological safety risk will have a compound amplification effect rather than a simple linear superposition. Therefore, the product term is used to characterize this type of coupled risk.

[0045] Step S4: Calculate the speed modulation coefficient based on the risk score.

[0046] In some embodiments, the velocity modulation coefficient is calculated based on the risk score, including: Establish a mapping function from the risk score to the target speed coefficient based on the risk score; The hysteresis velocity coefficient is obtained based on the hysteresis bandwidth and the mapping function; The constraint velocity coefficient is calculated based on the actual output velocity and hysteresis velocity coefficient of the previous moment. The output speed coefficient is calculated based on the constraint speed coefficient. The output speed coefficient is processed based on the indicator function to obtain the speed modulation coefficient.

[0047] Specifically, a mapping function from risk score to target speed coefficient is established based on risk assessment. This directly transforms the risk assessment results into a target benchmark for speed adjustment, establishing a correspondence between the speed adjustment magnitude and the risk level. Introducing hysteresis bandwidth to generate a hysteresis speed coefficient avoids frequent speed jumps caused by small fluctuations in the risk score, ensuring a smooth transition in output speed. Calculating the constraint speed coefficient based on the actual output speed from the previous moment ensures that speed changes follow a continuous trajectory, avoiding operational shocks caused by sudden speed changes. The speed modulation coefficient, obtained after processing the output speed coefficient by an indicator function, effectively limits the output range, ensuring that the final output speed adjustment parameters are within a reasonable range. Overall, this achieves smooth speed control driven by risk level, adapting to the speed adjustment needs of actual operational scenarios.

[0048] In one feasible implementation, a risk score is defined. Target speed coefficient Mapping function: ; in, To integrate risk scoring, The target speed coefficient (1.0 corresponds to full speed, and 0.3 is the minimum safe speed). Define a safe work area for low-risk thresholds. Define the danger zone boundary for the high-risk threshold. The physical meaning of the three-segment design is: low-risk zone ( ) Run at full speed to maximize efficiency, medium-risk area ( Linear deceleration balances safety and efficiency, decreasing from 100% to 50%, in high-risk areas. The speed is quickly reduced to the minimum safe speed of 30% to prioritize safety.

[0049] Furthermore, hysteresis logic is introduced to suppress frequent speed switching caused by small fluctuations in the risk score, resulting in a hysteresis speed coefficient. : ; in, and These are risk scores for the current moment and the previous moment, respectively. The actual output speed at the previous moment. The hysteresis bandwidth represents the speed at which the current speed is maintained when the risk change is less than 5%. This is a piecewise mapping function. Hysteresis is essentially a state machine memory mechanism; a transition is triggered only when the state change exceeds a threshold.

[0050] Furthermore, dynamic constraints are applied to the rate of change of velocity to ensure executability, resulting in the constraint velocity coefficient. : ; ; in, For the amplitude limiting function, The maximum deceleration rate (a maximum reduction of 10% in speed per 10ms control cycle, i.e., equivalent acceleration) ), Maximum acceleration rate (maximum speed increase of 5% per cycle, i.e.) The asymmetric design is based on the principle of safety first: rapid deceleration to protect the operator in dangerous situations, and slow acceleration to avoid being unable to adapt in time during recovery.

[0051] Step S5: Perform robot speed control operation based on the speed modulation coefficient.

[0052] In some embodiments, the speed modulation coefficient ranges between 0 and 1; When the speed modulation coefficient is 1.0, the robot runs at full speed; The robot stops when the speed modulation coefficient is 0.0.

[0053] By limiting the speed modulation coefficient to a range of 0 and 1, continuous stepless adjustment of the robot's speed from stop to full speed is achieved, creating a smooth correlation between speed changes and risk levels. When the speed modulation coefficient is 1.0, the robot operates at full speed; when it is 0.0, the robot stops. The control logic at both ends is clear, facilitating direct execution by the control system without additional interval calculations. Driving robot speed control with the speed modulation coefficient transforms the risk assessment results obtained from the physiological state into the robot's actual motion output, enabling dynamic speed regulation in human-robot collaborative scenarios. When risk increases, the speed is promptly reduced until the robot stops; when risk decreases, the speed gradually returns to normal operating speed, balancing safety during operation with overall operational efficiency.

[0054] In one feasible implementation, a hard distance threshold is set as the final safety barrier: ; in, This is an indicator function (returns 1 if the condition is true, otherwise returns 0). The Euclidean distance from the operator's torso center to the robot tool center point TCP is calculated in real time using a depth camera; cm is the safety distance threshold, calculated based on the SSM requirements of ISO 15066 standard, and taking into account robot braking distance and human reaction time. ;in, For maximum speed, s represents human reaction time. This is the robot's braking distance. This mechanism ensures that even if the risk scoring algorithm fails, the distance sensor can still trigger physical protection.

[0055] The physiological-based robot speed control method provided in the above embodiments can complete individual-based calibration through raw physiological signals and video stimulus intensity curves. Furthermore, by establishing a physiological response model and standardized mapping function, it can adapt to the differences in physiological responses among different operators, eliminating judgment bias caused by individual baseline differences. Real-time acquisition of multimodal signals input to the model and mapping function, and calculation of risk scores, allows for the simultaneous fusion of multi-dimensional information, ensuring that risk assessment aligns with the operator's actual situation. Continuous speed modulation coefficients output based on the risk score enable smooth transitions in robot speed adjustments, avoiding motion fluctuations caused by frequent state switching. This application can improve the comfort of the collaborative process and enhance the overall execution efficiency of robot operations while ensuring human-robot collaboration safety.

[0056] In one feasible implementation, the experiment uses a collaborative robotic arm with a lightweight gripper mounted at its end effector. The operator is positioned to the side and in front of the robot, performing parts delivery and assisting assembly tasks. The system collects the operator's pupil diameter P(t), the minimum distance d(t) between the robot's TCP and key points on the operator's torso / hand, the pupil signal quality Q_p(t), and the distance signal quality Q_d(t).

[0057] To ensure consistency with the aforementioned technical solution, the individualized calibration phase employs 60-second standardized visual stimuli, and calibration data is resampled at 10 Hz for parameter identification; the real-time control cycle is 10 ms. Distance measurements are in centimeters; example parameters are d_safe=100cm and d_stop=30cm.

[0058] After calibration, the real-time pupil signal is normalized to the standardized tension B(t) according to the individual response range: B(t)=clip((P(t)-P_0) / (P_{max}-P_0),0,1); Where P_0 is the individual relaxation baseline, and P_max is the upper limit of the high-tension response estimated by the response model during the calibration phase. Spatial proximity D(t) is defined as: D(t)=clip((d_{safe}-d(t)) / (d_{safe}-d_{stop}),0,1); Multimodal fusion follows the "quality weighting + interaction term" structure of the original technical solution, where the linear weighting term is: R_{lin}(t)=(Σ_i w_i Q_i(t)x_i(t)) / (Σ_i w_i Q_i(t)); In this embodiment, x_i(t) takes the values ​​of B(t) and D(t). The comprehensive risk score is: R(t)=clip(R_{lin}(t)+γB(t)D(t),0,1); Wherein, γB(t)D(t) is the interaction term, used to characterize the combined risk enhancement effect under "high tension and close proximity". The robot speed command is: v_{cmd}(t)=α(t)v_{ref}(t), α(t)∈[0,1]; Furthermore, the risk-speed mapping adopts the three-stage design of the above scheme: the low-risk zone runs at full speed, the medium-risk zone smoothly decreases from 100% to 50%, and the high-risk zone further decreases to the minimum safe speed of 30%.

[0059] α_{tar}(R)=1, R <R_{low}; α_{tar}(R)=1-0.5(R-R_{low}) / (R_{high}-R_{low}), R_{low}≤R <R_{high}; α_{tar}(R)=0.5-(0.5-α_{min})(R-R_{high}) / (1-R_{high}), R≥R_{high}; In this example, R_low=0.35, R_high=0.70, and α_min=0.30. The hysteresis bandwidth is set to h=0.05, and the previous output state is maintained when the risk score change is less than this bandwidth. The speed change rate constraint is the same as the original scheme: the maximum deceleration is 0.10 and the maximum acceleration is 0.05 within every 10 ms control cycle.

[0060] α_k=clip(α_{tar,k},α_{k-1}-0.10,α_{k-1}+0.05); Distance hard threshold protection maintains the original scheme's role as a final safety barrier: d(t)≤d_{stop}⇒ α(t)=0.

[0061] Furthermore, the experimental evaluation indicators are set around the technical effects of the above embodiments, including early deceleration, suppression of peak psychological stress, speed smoothness, efficiency maintenance, and sensor abnormal degradation capability.

[0062] ΔT_{pre}=t_{stop}-t_{α<0.95}; J_α=(1 / (N-2))Σ_{k=3}^{N}|α_k-2α_{k-1}+α_{k-2}|; Where ΔT_pre represents the pre-deceleration time relative to the hard threshold trigger time; J_α is the second-order difference mean of the velocity modulation coefficient, used to quantify velocity jitter. The number of psychological stress peaks N_stress is defined as the number of events where B(t) > 0.75 and lasts for more than 0.5 s.

[0063] In some embodiments, differences exist in P_0, P_max, and response delay τ among operators, indicating that directly using a uniform pupil threshold is not suitable for cross-individual comparisons. The calibration R² values ​​are all above 0.80, meeting the calibration quality requirements of the example experiment.

[0064] In some embodiments, during a collaborative approach task, the robot's end effector approaches the operator's work area from a distance of 120 cm, maintains proximity for approximately 20 seconds, and then withdraws. Throughout the normal approach, d(t) does not decrease less than d_stop, thus not triggering an emergency stop; this setting is used to observe whether the psycho-spatial fusion risk can be mitigated by adjusting the speed smoothly in advance before a hard threshold.

[0065] Please see Figure 2 Another embodiment of this application provides a physiologically based robot speed control device, the device comprising: The acquisition module 101 is used to acquire raw physiological signals and video stimulus intensity curves.

[0066] Module 102 is used to build an individual physiological response model and a standardized mapping function based on the original physiological signals and video stimulus intensity curves.

[0067] The calculation module 103 is used to acquire real-time multimodal signals, input the real-time multimodal signals into an individual physiological response model and a standardized mapping function, and calculate to obtain a risk score.

[0068] The modulation module 104 is used to calculate the speed modulation coefficient based on the risk score.

[0069] The control module 105 is used to perform robot speed control operations based on the speed modulation coefficient.

[0070] The specific limitations of the physiological-based robot speed control device provided in this embodiment can be found in the embodiment of the physiological-based robot speed control method described above, and will not be repeated here. Each module in the above-described physiological-based robot speed control device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0071] This application provides a computer device that may include a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface communicates with external terminals via a network connection. When the computer program is executed by the processor, it causes the processor to perform the steps of a physiologically based robot speed control method as described in any of the above embodiments.

[0072] The working process, working details, and technical effects of the computer device provided in this embodiment can be found in the embodiment of a physiological-based robot speed control method described above, and will not be repeated here.

[0073] This application provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the steps of a physiological-based robot speed control method as described in any of the above embodiments. The computer-readable storage medium refers to a data storage medium, which may include, but is not limited to, floppy disks, optical disks, hard disks, flash memory, USB flash drives, and / or memory sticks. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0074] The working process, working details, and technical effects of the computer-readable storage medium provided in this embodiment can be found in the embodiment of a physiological-based robot speed control method described above, and will not be repeated here.

[0075] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0076] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0077] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A physiological-based robot speed control method, characterized in that, The method includes: Acquire raw physiological signals and video stimulus intensity curves; Based on the original physiological signals and the video stimulus intensity curve, an individual physiological response model and a standardized mapping function are established. Real-time multimodal signals are acquired, and the real-time multimodal signals are input into the individual physiological response model and standardized mapping function for calculation to obtain a risk score; The speed modulation coefficient is calculated based on the risk score. The robot speed is controlled based on the speed modulation coefficient.

2. The physiological-based robot speed control method according to claim 1, characterized in that, The step of establishing an individual physiological response model and a standardized mapping function based on the original physiological signals and the video stimulus intensity curve includes: A standardized mapping function is calculated based on the original physiological signals, the relaxation baseline value, and the tension upper limit value. An individual physiological response model is established based on the video stimulus intensity curve, gain coefficient, relaxation baseline value, and physiological response delay.

3. The physiological-based robot speed control method according to claim 1, characterized in that, The process of acquiring real-time multimodal signals, inputting these signals into the individual physiological response model and standardized mapping function, and calculating the risk score includes: Acquire real-time multimodal signals; the real-time multimodal signals include standardized physiological signals, spatial signals, and signal quality coefficients; The real-time multimodal signal is input into the individual physiological response model and the standardized mapping function to perform time delay estimation and nonlinear least squares fitting to obtain the model prediction value; A risk score is calculated based on the average of the measured data and the predicted value of the model.

4. The physiological-based robot speed control method according to claim 1, characterized in that, The calculation of the speed modulation coefficient based on the risk score includes: A mapping function from the risk score to the target speed coefficient is established based on the risk score. The hysteresis velocity coefficient is obtained based on the hysteresis bandwidth and the mapping function. The constraint velocity coefficient is calculated based on the actual output velocity at the previous moment and the hysteresis velocity coefficient. The output speed coefficient is calculated based on the constrained speed coefficient. The output speed coefficient is processed based on the indicator function to obtain the speed modulation coefficient.

5. The physiological-based robot speed control method according to claim 1, characterized in that, The original physiological signal is the pupil diameter detected when the operator watches a standardized stimulation video of a preset duration.

6. The physiological-based robot speed control method according to claim 1, characterized in that, The method further includes: Normalized visual motion velocity was calculated based on optical flow. The normalized acceleration is obtained based on the velocity difference between adjacent frames; The video stimulus intensity curve is constructed based on the weighting coefficients, normalized height, normalized visual motion velocity, and normalized acceleration.

7. The physiological-based robot speed control method according to claim 4, characterized in that, The speed modulation coefficient is in the range of 0 and 1; When the speed modulation coefficient is 1.0, the robot runs at full speed; The robot stops when the speed modulation coefficient is 0.

0.

8. A physiologically based robot speed control device, characterized in that, The device includes: The acquisition module is used to acquire raw physiological signals and video stimulus intensity curves; The module is used to build an individual physiological response model and a standardized mapping function based on the original physiological signals and the video stimulus intensity curve; The calculation module is used to acquire real-time multimodal signals, input the real-time multimodal signals into the individual physiological response model and standardized mapping function, and calculate to obtain a risk score; A modulation module is used to calculate the speed modulation coefficient based on the risk score; The control module is used to perform robot speed control operations based on the speed modulation coefficient.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the physiological-based robot speed control method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the physiological-based robot speed control method as described in any one of claims 1 to 7.

Images