Seat pre-adjustment control method based on embedded CNN waist-hip trajectory dynamic tracking

Abstract

The invention discloses a seat pre-adjustment control method based on embedded CNN waist-hip trajectory dynamic tracking, and the method can achieve the precise prediction of the motion intentions of people with different heights through the construction of a height proportion change curve, effectively eliminates the motion recognition error caused by the height and body type difference of different individuals, and improves the accuracy of the motion recognition. The method enhances the refinement of seat control, and reflects the change of the action stage of the target person by setting the track slope, thereby recognizing the transition of the target person from sitting down to stable sitting posture and from sitting posture to getting up in different action stages, predicting the next trend of the action in advance, effectively improving the prediction accuracy of the action intention, and improving the user experience. The seat system can make a response when the action starts, and the situation that the comfort is affected due to delayed adjustment is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of seat intelligent control, and in particular to a seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN. Background Art

[0002] With the development of smart home and smart cockpit systems, the automatic adjustment function of seats has become an important technical means to improve user experience and enhance ergonomic adaptability.

[0003] Existing seat assist systems are mostly based on static position sensing or trigger-based responses. These systems typically utilize pressure sensors or motion switches to detect whether a user is sitting down or preparing to stand up, before initiating simple lifting or tilting movements. These approaches fail to proactively identify user intent and only react after the user has completed key movements or made contact with the seat. This results in delayed interaction and a lack of ability to anticipate and adapt to human movements, severely impacting the user experience. This is particularly pronounced for individuals with limited physical control, such as the elderly and rehabilitation patients.

[0004] Therefore, it is necessary to design a seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN to solve the above problems. Summary of the Invention

[0005] The present invention overcomes the shortcomings of the prior art and provides a seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN.

[0006] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is: a seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN, comprising the following steps:

[0007] Step S1: real-time acquisition of three-dimensional position coordinate information of key points of a target person within a target range to form a sequence of human body trajectory points;

[0008] Step S2: Based on the human body trajectory point sequence, analyze the change trend of the target person's height ratio over time in real time to form a height ratio change curve;

[0009] Step S3: Based on the height ratio change curve, determine the target person's current action stage and predict the target person's action intention;

[0010] Step S4: Dynamically adjust the seat according to the predicted target person's movement intention.

[0011] In a preferred embodiment of the present invention, in step S1, the key points include: the head, neck, waist, hips and lower limbs.

[0012] In a preferred embodiment of the present invention, in step S1, the key points of the target person are set as in, Represents the three-dimensional coordinates of the nth key point in time period t;

[0013] The nth key point is collected in real time with a fixed sampling period to form a trajectory point sequence of the nth key point The trajectory point sequences of several key points are aggregated to form a human body trajectory point sequence.

[0014] In a preferred embodiment of the present invention, step S2 includes the following sub-steps:

[0015] Step S21: obtaining the initial height of the target person;

[0016] Step S22: Based on the human body trajectory point sequence, select the key points of the target person's vertical height correlation, obtain the target person's height in different time periods, and calculate the height ratio The value range of R(t) is [0, 1], zt represents the height of the target person at the tth moment, and H represents the initial height of the target person;

[0017] Step S23: Sample and cache R(t) at time intervals to obtain a continuous height ratio change sequence R={R(t0), R(t1), ..., R(t n )}.

[0018] In a preferred embodiment of the present invention, in step S22, the key points of the target person's longitudinal height correlation degree are selected including: head, neck, and waist.

[0019] In a preferred embodiment of the present invention, in step S3, the action stage identification includes but is not limited to: approach stage, sitting transition stage, sitting buffer stage, stable sitting stage, standing up preparation stage, waist lifting and leaving stage, and standing up completion stage.

[0020] In a preferred embodiment of the present invention, in step S3, corresponding height ratio threshold intervals are set for different motion stages according to different motion stages and height ratios of the human body.

[0021] In a preferred embodiment of the present invention, in step S3, the trajectory slope of the height ratio R′(t)≈R(t)-R(t-1) is calculated, and combined with the height ratio threshold interval of the target person, the next action intention of the target person is predicted according to the change direction of R′(t).

[0022] In a preferred embodiment of the present invention, in step S3, predicting the target person's next action intention requires determining whether the current intention is stable, including the following steps:

[0023] Step S31: Acquire characteristic parameters of the target person's height ratio, waist-hip angle, key point speed, and trajectory slope;

[0024] Step S32: Score the actions at different times according to the characteristic parameters Where wi represents the weight of the i-th feature, and fi(t) represents the score of the i-th feature;

[0025] Step S33: Make a judgment based on the set action intention threshold. If S(t) is greater than the action intention threshold, the current action is judged to be in the action stage and the action intention is accurate. If S(t) is less than the action intention threshold, the current action is judged to be in the action stage and the action intention is inaccurate, and the current action state is maintained.

[0026] The present invention solves the defects existing in the background technology and has the following beneficial effects:

[0027] (1) The present invention provides a seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN. By real-time acquisition of a three-dimensional posture trajectory point sequence formed by key points such as the head, neck, and waist, and calculating and normalizing the obtained height ratio change curve, the method can effectively reflect the height changes of the longitudinal center of gravity of the human body. In combination with the trajectory slope and the current height ratio interval, the method can accurately identify the stage of the target person, accurately judge the current action intention of the target person, and realize dynamic transition prediction from the current state to the next state, thereby accurately controlling the adjustment of the seat, ensuring that the seat is always in the best state, improving the seat usage experience and comfort, and thus realizing personalized and intelligent seat control.

[0028] (2) The present invention provides a seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN. By constructing a height ratio change curve, the method can accurately predict the movement intentions of people of different heights, effectively eliminate the movement recognition errors caused by differences in height and body shape among individuals, and enhance the refinement of seat control. At the same time, by setting the trajectory slope to reflect the changes in the movement phases of the target person, the method can identify the transition of the target person from sitting down to a stable sitting position, from a sitting position to standing up, and other different movement phases, thereby predicting the next trend of the movement in advance, effectively improving the prediction accuracy of the movement intention, and allowing the seat system to respond at the beginning of the movement to avoid affecting comfort due to delayed adjustment.

[0029] (3) The present invention provides a seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN. The method determines the stability of intentions through scoring and thresholds. A staged scoring mechanism is used to determine the stability of different action stages for the height ratio fluctuations caused by natural posture adjustments such as neck deflection and waist bending during the stable sitting posture. This avoids the influence of height ratio fluctuations caused by natural posture adjustments such as neck deflection and waist bending on the prediction of action intentions, thereby effectively avoiding misjudgments caused by single height ratio changes, significantly improving the recognition accuracy of real action intentions, and ensuring stability in the face of complex action transitions or body posture changes in action intention prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments described in the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive efforts.

[0031] Figure 1 This is a flow chart of the seat pre-adjustment control method based on the dynamic tracking of waist and hip trajectories by the embedded CNN of the present invention. DETAILED DESCRIPTION

[0032] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0033] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0034] In the description of the present application, it should be understood that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the scope of protection of the present application. In addition, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, features defined as "first", "second", etc. may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0035] In the description of this application, it should be noted that, unless otherwise expressly specified or limited, the terms "installed," "connected," and "connected" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; and they can refer to internal connections between two components. Those skilled in the art will understand the specific meanings of the above terms in this application based on specific circumstances.

[0036] like Figure 1 As shown, the technical solution adopted by the present invention is: a seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN, comprising the following steps:

[0037] Step S1: real-time acquisition of three-dimensional position coordinate information of key points of a target person within a target range to form a sequence of human body trajectory points;

[0038] Step S2: Based on the human body trajectory point sequence, analyze the change trend of the target person's height ratio over time in real time to form a height ratio change curve;

[0039] Step S3: Based on the height ratio change curve, determine the target person's current action stage and predict the target person's action intention;

[0040] Step S4: Dynamically adjust the seat according to the predicted target person's movement intention.

[0041] In the present invention, in step S1, the key points include: the head, neck, waist, buttocks and lower limbs.

[0042] In the present invention, in step S1, the key points of the target person are set as in, Represents the three-dimensional coordinates of the nth key point in time period t;

[0043] The nth key point is collected in real time with a fixed sampling period to form a trajectory point sequence of the nth key point The trajectory point sequences of several key points are aggregated to form a human body trajectory point sequence.

[0044] At least one camera is pre-set around the seat to ensure that there is no obstruction or strong backlight within the user's normal sitting position. The camera is calibrated with internal and external parameters to ensure that the acquired depth map or binocular disparity map has good 3D reconstruction accuracy.

[0045] A lightweight embedded CNN pose estimation algorithm is deployed in an embedded platform with limited local computing power. The model input is a single-frame RGB image, and the output is a set of 2D pixel coordinates of human key points and their confidence levels, which are converted into corresponding 3D spatial coordinates based on depth information or binocular parallax.

[0046] In the present invention, step S2 includes the following sub-steps:

[0047] Step S21: obtaining the initial height of the target person;

[0048] Step S22: Based on the human body trajectory point sequence, select the key points of the target person's vertical height correlation, obtain the target person's height in different time periods, and calculate the height ratio The value range of R(t) is [0, 1], zt represents the height of the target person at the tth moment, and H represents the initial height of the target person;

[0049] Step S23: Sample and cache R(t) at time intervals to obtain a continuous height ratio change sequence R={R(t0), R(t1), ..., R(t n )}.

[0050] In the present invention, in step S22, the key points of the target person's longitudinal height correlation degree are selected to include: head, neck, and waist.

[0051] In seat pre-adjustment control, the target person's sitting or standing movements are often accompanied by vertical movement of the body's center of gravity. Directly judging the movement based on the "absolute coordinates" of key points in three-dimensional space will be affected by individual body shape differences, camera viewing angles, or changes in installation height. Therefore, the system first collects representative key points of the human body, such as the head, neck, waist, and hips. Using the height of the initial standing posture as the standard, it calculates the proportion of the current human height under this standard. This proportion is called the height ratio R(t), and its value range is [0,1]. Through standardization, the action recognition model no longer relies on the individual's actual height, but instead reflects the action trend through relative proportions.

[0052] To accurately capture height changes, key points along the central axis of the human body, such as the head, neck, and waist, are selected, and the vertical height difference of these positions in three-dimensional space is extracted to calculate the current visible height. This value is compared with the initial initialized standard height to obtain the standardized height ratio. This ratio value is cached in each time frame to form a continuous sequence curve that changes over time. The height ratio curve maps the continuous state changes of the human body during sitting down or standing up. For example, a rapid decline represents the transition to sitting down, a stable low position indicates maintaining a sitting posture, and a slow rise may indicate the beginning of standing up.

[0053] It should be noted that the height ratio change curve is structured data in the form of a time series, which not only contains the instantaneous state at a certain moment, but also reflects the trend information of the evolution of human posture. By calculating the slope of the curve in a short time window, it can be judged whether the target person is in a static state (such as sitting), performing a movement transfer (such as raising the waist or sitting down), or about to perform a new action (such as preparing to get up). The height ratio change curve can not only assist in the judgment of the current action stage, but also lay the foundation for the prediction of the next action. When the curve is continuously in a certain interval and shows a specific trend (such as from stable to rising), the system can predict that the target person is about to stand up and adjust the seat height or inclination in advance to achieve "intention prediction" control. This method has fast response and small computational complexity, and is suitable for edge computing devices or embedded intelligent systems.

[0054] In the present invention, in step S3, the action phase identification includes but is not limited to: approach phase, sitting transition phase, sitting buffer phase, stable sitting phase, standing up preparation phase, waist lifting phase, and standing up completion phase;

[0055] In step S3, corresponding height ratio threshold intervals are set for different movement stages according to different movement stages and height ratios of the human body.

[0056] In step S3, the trajectory slope of the height ratio R′(t)≈R(t)-R(t-1) is calculated, combined with the height ratio threshold interval of the target person, which is set based on experience, and the target person's next action intention is predicted according to the change direction of R′(t).

[0057] Action phase status table:

[0058] stage Height ratio threshold interval trajectory slope Action Judgment Approaching stage 1.0→0.95 ≈0 Close to the seat, but not yet seated Sit-down transition 0.95→0.7 Less than 0, continuous The body sinks quickly and starts to sit down Seat cushioning 0.7→0.6 Less than 0, continuous Body-touching seat, cushioned Stable sitting posture 0.6→0.4 ≈0 Complete sitting and maintain a stable posture Get up and get ready 0.4→0.5 Greater than 0, continuous Prepare to stand up Lift your waist and leave your seat 0.6→0.9 Greater than 0, continuous Get up and leave quickly Stand up and finish >0.95 ≈0 Get out of your seat completely and stand

[0059] In the present invention, in step S3, predicting the target person's next action intention requires determining whether the current intention is stable, including the following steps:

[0060] Step S31: Acquire characteristic parameters of the target person's height ratio, waist-hip angle, key point speed, and trajectory slope;

[0061] Step S32: Score the actions at different times according to the characteristic parameters Where wi represents the weight of the i-th feature, and fi(t) represents the score of the i-th feature;

[0062] Step S33: Make a judgment based on the set action intention threshold. The action intention threshold is obtained based on experience and is preferably set to 0.85. If S(t) is greater than the action intention threshold, the current action is judged to be in the action stage and the action intention is accurate. If S(t) is less than the action intention threshold, the current action is judged to be in the action stage and the action intention is inaccurate, and the current action state is maintained.

[0063] Each feature parameter has a corresponding action intention feature value range. If it falls within the range, the score is high, otherwise the score is low. The format is as follows:

[0064]

[0065] fi(t) represents the i-th feature, [Li,Ui] represents the normal interval of the feature required in the action phase, and δ represents the relaxation tolerance range, which is defined based on experience.

[0066] In order to achieve active adaptation and advance response of the seat control to the user's behavior, it is necessary to clarify which stage the target person's current body movement is in. By obtaining the target person's height ratio change curve, the ratio change and its change rate can be used to determine the stage of their dynamic behavior. For example, when a person changes from standing to sitting, the height ratio will slowly or quickly drop from close to 1 to about 0.6, and show specific trend characteristics at different stages. Based on these characteristics, the height ratio threshold interval and trajectory slope change rules corresponding to each stage are set, and then the action stages such as "approaching", "sitting down transition", and "stable sitting posture" are accurately identified;

[0067] After accurately identifying the action phase, the system further determines the target person's next behavioral trend, predicting their action intention. For example, when a person is in the "stable sitting position" phase and their height ratio is slowly increasing, combined with characteristics such as a decreasing waist-hip angle and changes in waist acceleration, it predicts that the user may be entering the "preparing to stand up" phase. The key here is not just observing the current value of the height ratio, but also combining its changing slope with relevant body parameters to predict the target person's behavioral intention through time series analysis and trend judgment. This prediction method can anticipate the target action before it occurs, enabling preemptive adjustment of the seat state.

[0068] It should be noted that when the target person sits on the seat, the deflection of the neck and the bending of the waist will cause the height ratio to change, which will affect the accuracy of their judgment and prediction. In order to avoid misjudging the intention due to temporary movement fluctuations, the intention stability judgment based on the scoring mechanism is introduced. Specifically, within the continuous time window, multiple features are quantified and scored, and the comprehensive intention score is calculated by weight. Only when the score continues to exceed the set threshold, the intention is judged to be "stable and reliable" and the seat control command is executed accordingly. This effectively improves the system's tolerance to changes in human micro-movements, avoids misadjustments, and enhances the reliability of system response and user experience.

[0069] In step S4, the seat is dynamically adjusted according to the predicted target person's movement intention;

[0070] Specifically, when step S3 determines that the user is in the "sitting transition" stage and the scoring result exceeds the sitting intention threshold, the seat is controlled to enter the "seat lowering" mode, and the current seat height and the preset target sitting height are queried. The preset target sitting height is calculated based on the user's initial height;

[0071] The optimal descent speed and acceleration / deceleration curves are dynamically calculated based on the user's current height ratio slope and key point speed, ensuring that the seat descends in accordance with the user's natural falling rhythm while avoiding sudden "bang" impacts.

[0072] When real-time monitoring detects that the user's height ratio remains within the "stable sitting posture" threshold range and the seat height is close to the preset target sitting height, the seat is controlled to automatically exit the "lowering mode" and switch to the "stable sitting posture" mode, locking the seat height and maintaining the current posture without adjustment. The seat can not only synchronize with the human body's movement state during the user's sitting process, but also achieve dynamic coordination in multiple links such as triggering intention, speed planning, execution drive and safety closed loop, which not only ensures user safety but also provides a seamless and comfortable experience, effectively assisting the target person in sitting down.

[0073] It should be noted that the data in the above steps S1 to S4 are all normalized.

[0074] The above description is based on the ideal embodiment of the present invention. Based on the above description, relevant personnel can make various changes and modifications without departing from the technical scope of this invention. The technical scope of this invention is not limited to the content of the specification and must be determined according to the scope of the claims.

Claims

1. A seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN is characterized by: The following steps are involved: Step S1: real-time acquisition of three-dimensional position coordinate information of key points of a target person within a target range to form a sequence of human body trajectory points; Step S2: Based on the human body trajectory point sequence, analyze the change trend of the target person's height ratio over time in real time to form a height ratio change curve; Step S3: Based on the height ratio change curve, determine the target person's current action stage and predict the target person's action intention; Step S4: Dynamically adjust the seat according to the predicted target person's movement intention.

2. The seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN according to claim 1 is characterized in that: In step S1, key points include: head, neck, waist, hips and lower limbs.

3. The seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN according to claim 2, characterized in that: In step S1, the key points of the target person are set as in, Represents the three-dimensional coordinates of the nth key point in time period t; The nth key point is collected in real time with a fixed sampling period to form a trajectory point sequence of the nth key point The trajectory point sequences of several key points are aggregated to form a human body trajectory point sequence.

4. The seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN according to claim 2, characterized in that: In step S2, the following sub-steps are included: Step S21: obtaining the initial height of the target person; Step S22: Based on the human body trajectory point sequence, select the key points of the target person's vertical height correlation, obtain the target person's height in different time periods, and calculate the height ratio The value range of R(t) is [0, 1], zt represents the height of the target person at the tth moment, and H represents the initial height of the target person; Step S23: Sample and cache R(t) at time intervals to obtain a continuous height ratio change sequence R={R(t0), R(t1), ..., R(t n )}.

5. The seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN according to claim 4 is characterized in that: In step S22, the key points of the target person's longitudinal height correlation degree are selected to include: head, neck, and waist.

6. The seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN according to claim 1 is characterized in that: In step S3, the action stage identification includes but is not limited to: approach stage, sitting transition stage, sitting buffer stage, stable sitting posture stage, standing up preparation stage, waist lifting and leaving stage, and standing up completion stage.

7. The seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN according to claim 6 is characterized in that: In step S3, corresponding height ratio threshold intervals are set for different movement stages according to different movement stages and height ratios of the human body.

8. The seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN according to claim 7 is characterized in that: In step S3, the trajectory slope of the height ratio R′(t)≈R(t)-R(t-1) is calculated, and combined with the height ratio threshold interval of the target person, the next action intention of the target person is predicted according to the change direction of R′(t).

9. The seat pre-adjustment control method based on dynamic tracking of waist and hip trajectories using an embedded CNN according to claim 8, characterized in that: In step S3, predicting the target person's next action intention requires determining whether the current intention is stable, including the following steps: Step S31: Acquire characteristic parameters of the target person's height ratio, waist-hip angle, key point speed, and trajectory slope; Step S32: Score the actions at different times according to the characteristic parameters Where wi represents the weight of the i-th feature, and fi(t) represents the score of the i-th feature; Step S33: Make a judgment based on the set action intention threshold. If S(t) is greater than the action intention threshold, the current action is judged to be in the action stage and the action intention is accurate. If S(t) is less than the action intention threshold, the current action is judged to be in the action stage and the action intention is inaccurate, and the current action state is maintained.

Images