g 按照坐标变换的标准记法，定义一组参数 RPY角（^α，Α ，表示 — ^^Ζ坐标系先绕其_Ζ轴转，再绕其 F轴 ^，最后绕其 Z轴; 得到。这样得到由固定坐标系 o一^Xj^Yj^Zj到⁰—^X'^Yi^Zi的变换矩阵的另一个表示如下。于是同样可以得到一个方程组，求解这个方程可以得到加速度计姿态。但这个方程组的解有多组。当加速度计初始姿态和最终姿态都有多解时，问题变得很复杂。因此，简化问题，固定初始姿态。假设为水平放置或竖直放置。g According to the standard notation of coordinate transformation, define a set of parameters RPY angle (^α , Α, denote - ^^Ζ coordinate system first around its_Ζ axis, then around its F axis ^, and finally around its Z axis; get. Transformation from a fixed coordinate system o -^X j^Y j^Z j to⁰ -^X '^Y i^Z i Another representation of the matrix is as follows. So you can also get a system of equations, and solve this equation to get the accelerometer pose. But there are many sets of solutions for this system of equations. When the initial attitude and final pose of the accelerometer have multiple solutions, the problem becomes complicated. Therefore, the problem is simplified and the initial pose is fixed. It is assumed to be placed horizontally or vertically.

另外，也可以认为从⁰—^xjYj^Zj到 o— i的变换是绕In addition, it can also be considered that the transformation from⁰ —^x jYj^Zj to o — i is

X γζ k = (k_v, k„, k_X γζ k = (k_v , k„, k_

下的一个固定向量' - y^ -y--= ，转动角得到，变换矩阵^ ^的另一种表示方法为： k_xk_xv rs(0)sk{<9)+cos(0) k_yk_xvefs{6) ~ sin( k_:k_xvers{0) +k_y svc(9) 风 Κ,θ)= k_xk_yvers(0)+k_:s 9) k_yk_yvers(0)+coe e) k,k_yvers{e)-k_xsvc e)

当已知初始姿态时，矩阵方程求解后的最终姿态存在多解。此时The next fixed vector ' - y^ -y--= , the rotation angle is obtained, and another representation of the transformation matrix ^ ^ is: k_x k_x v rs(0)sk{<9)+cos(0) k_y k_x vefs{6) ~ sin( k_: k_x vers{0) +k_y svc(9) 风Κ,θ)= k_x k_y vers(0)+k_: s 9) k_y k_y Vers(0)+coe e) k,k_y vers{e)-k_x svc e)

When the initial pose is known, there are multiple solutions to the final pose after the matrix equation is solved. at this time

(¾， ) = 1, 2· · . (3⁄4, ) = 1, 2· · .

个最终姿态，都能找到一个对应的 a final gesture, you can find a corresponding

令：

。则加速计由初始姿态经过变变换换得得到到最最终终姿姿态态。。这这样样可可以以认认为为整整个个过过程程中中，，变变换换都都是是绕绕"^j 轴进行。于是整个转动过程可以差分成很多绕^'进行的一连串小角度make:

. Then, the accelerometer is converted from the initial attitude to the final final attitude state. . This can be considered as a whole process, and the transformation is performed around the "^j axis. So the whole rotation process can be differentiated into a series of small angles around the ^'

算法如下：加速度计的姿态用固联在加速度计上的坐标系表示，假设初始姿态为⁷^ 终止姿态为。示意图参见附图 6。从图 6 可以看出，初始时刻测量的加速度矢量为 T N2009/000770The algorithm is as follows: The attitude of the accelerometer is represented by the coordinate system fixed on the accelerometer, assuming that the initial pose is⁷ ^ and the end pose is . See Figure 6 for a schematic diagram. As can be seen from Figure 6, the acceleration vector measured at the initial moment is T N2009/000770

。终止时刻测量的加速度矢量为 lxN '^ayN a. N. The acceleration vector measured at the end time is lxN '^a yN a. N

。假设初始姿态到终止姿态 ^的变换矩阵为 . Suppose the transformation matrix from initial pose to termination pose ^ is

，终止姿态^到初始姿态的变换矩阵为贝 IJ:, the transformation matrix of the termination pose ^ to the initial pose is the shell IJ:

N On-1N On-1

0^{R =} N^R0^{R =} N^R

SN二

假设终止姿态变换到初始姿态的 RPY角为（^α，，则 cacfi,c asfisy一 sacy, c as cy + sasy

sa c β, sasfisy + cacy, sasficy一 easySN II

Assume that the RPY angle of the termination pose to the initial pose is (^α , , then cacfi, c asfisy - sacy, c as cy + sasy

Sa c β, sasfisy + cacy, sasficy an easy

得方程组-

Equations -

此方程组有多解，针对方程组的每一组（^tt，A 代入姿态矩阵 '^R, 对其求逆，得到矩阵。 70 假设由初始姿态^。变换到姿态等效于绕坐标系内固定单位向

This system of equations has multiple solutions. For each group of equations (^tt , A is substituted into the attitude matrix '^R , invert it to get the matrix. 70 assumed by the initial pose ^. Transform to pose is equivalent to a fixed unit in the coordinate system

=(k k k ) n=(k k k ) n

λ"^y'^j 转动 ^角得到，贝 IJ

λ"^y '^j turns ^ corner to get, Bei IJ

则得到如下方程组: Then get the following equations:

n_x +o y +a_zz = (k_x + k + k_z )vers6 + 3c0

于是得到- sin<9 =士^ ~a_yf +{a_x ~n_zf + {n_y-oJn_x +oy +a_z z = (k_x + k + k_z )vers6 + 3c0

Then get - sin<9 =士^~a_y f +{a_x ~n_z f + {n_y -oJ

k 二。:_k 二 ―":_{k =}ⁿy -°_X

k II. :_k二―":_{k =}ⁿ y -°_X

x 2sin6> '^y 2 sin (9 ' '" 2sin^ 由于方程存在病态解情况，实际中系统要求偏离 0和 180°，这是因为方程此时无法准确快速收敛。x 2sin6>'^y 2 sin (9 ''" 2sin^ Due to the ill-conditioned solution of the equation, the actual system requirements deviate from 0 and 180°, because the equation cannot be accurately and quickly converge at this time.

方程组（1) 的解有多组，对每一组解都能求出唯一的一组^，）。则^^=Ι^^η(6>')，对应的存在（，）。设加速度计在每次采样得到的加速度信息中，重力加速度的影响值依次为⁼(^/，^''^-^_)^{= ()} '·^Ν。每个釆样点时，加速度计的姿态为⁷^ '^{= ()}，²'''^，从⁷^到初始姿态⁷ ^的变换矩阵为。^。显然There are multiple sets of solutions for equation (1), and a unique set of ^, ) can be obtained for each set of solutions. Then ^^{= Ι} ^^η(6> '), the corresponding existence ( , ). In the acceleration information obtained by each accelerometer, the influence value of gravity acceleration is⁼ (^/ , ^''^-^_ )^{= ()} '·^Ν . For each sample point, the attitude of the accelerometer is⁷ ^ '^{= ()} ,² '''^, and the transformation matrix from⁷ ^ to the initial pose of⁷ ^ is . ^. Obviously

₀⁰R = /(单位阵)。则 &二 o^R~^lSo。₀⁰ R = / (unit array). Then & two o^R ~^l So.

利用估计每个变换矩阵。' ，这里采用的方法是认定

在每个釆样点，采集到的加速度为

Use to estimate each transformation matrix. ' , the method used here is to identify

At each sample point, the acceleration is

则去除重力在各轴上产生的影响后，运动加速度为二^一&。此时的 A '是在姿态 η下的信息，需要转化到初始姿态 r_fl下，则在每个采样点，在初始姿态下的加速度信息为-Then, after removing the influence of gravity on each axis, the motion acceleration is two ^ one. At this time, A ' is the information under the attitude η, and needs to be converted to the initial attitude r_fl . At each sampling point, the acceleration information in the initial attitude is -

= ( =_Q!R («/ 一 g i ) = 0^{R a}i ― g o 如图 7所示，来自静止-运动检测模块的完整运动数据在重力加速度分离模块中被两次使用。首先被用于计算运动的最小转角以及对应的转轴方向，得到每一个数据采样点的姿态转换矩阵之后利用 ^ 计算重力加速度在 X/Y/Z轴的分量，从来自静止-运动检测模块的完整运动数据中每一个釆样点数据 [a_x,a_y,a_z]中分离，得到实际完整人体运动过程中的加速度数据，最后利用实际完整人体运动过程中的加速度数据，经过一次积分计算出运动速度，再经过一次积分计算出运动轨迹。同时利用 ^计算重力加速度在 X/Y/Z轴的分量同地心引力产生的重力加速度的分布情况，计算出每一个数据采样点时刻设备的空间姿态。最终实际完整人体运动过程中的加速度、速度、位移，连同设备的空间姿态，共同组成了完整运动数据。算法（二）：= ( =_Q !R («/ a gi ) = 0^{R a} i ― go As shown in Figure 7, the complete motion data from the stationary-motion detection module is used twice in the gravity acceleration separation module. Calculate the minimum rotation angle of the motion and the corresponding rotation axis direction, obtain the attitude transformation matrix of each data sampling point and calculate the component of the gravity acceleration in the X/Y/Z axis by ^, from the complete motion data from the static-motion detection module. A sample point data [a_x , a_y , a_z ] is separated to obtain the acceleration data of the actual complete human motion process. Finally, the acceleration data in the actual complete human motion process is used to calculate the motion speed after one integral, and then After one integral, the motion trajectory is calculated. At the same time, the distribution of the gravity acceleration generated by the gravity acceleration in the X/Y/Z axis and the gravity of the gravity is calculated, and the spatial attitude of the device at each data sampling point is calculated. The acceleration, velocity, and displacement of the complete human motion, together with the spatial attitude of the device, together form the complete motion data. Algorithm (2):

除了上述算法（一）夕卜，也可以通过其他算法将重力加速度从传感器釆集的数据中分离。通过静止-运动检测模块检测到运动幵始后，首先将运动开始之前的静止数据作为初始化的第一组静止姿态 μ1，并用这个姿态 μΐ的传感器数据，作为重力加速度 gl在姿态 μΐ时在 x/y/z轴的分量 (gxl,gyl，gzl)，并把接下来较短固定时间内的三轴加速度数据 (axi,ayi，azi)去除重力加速度 gl此时的 x/y/z轴的分量 (gxl,gyl,gzl)， In addition to the above algorithm (1), gravity acceleration can also be separated from the data collected by the sensor by other algorithms. After detecting the start of the motion by the stationary-motion detecting module, the static data before the start of the motion is first used as the initial set of the stationary pose μ1, and the sensor data of the pose μΐ is used as the gravity acceleration gl at the attitude μΐ at x/ The components of the y/z axis (gxl, gyl, gzl), and remove the triaxial acceleration data (axi, ayi, azi) of the next shorter fixed time to remove the component of the x/y/z axis of the gravitational acceleration gl at this time. (gxl, gyl, gzl),

(Axi,Ayi,Azi) = (axi,ayi,azi) - (gxl,gyl,gzl) (Axi, Ayi, Azi) = (axi,ayi,azi) - (gxl,gyl,gzl)

(Axi,Ayi,Azi)是这段时间内的运动加速度数值,计算出运动速度和轨迹，并与模式匹配数据库当中的有限组运动指令进行匹配，得出一组确定的运动方向 al(kxl,kyl,kzl)。 (Axi, Ayi, Azi) is the motion acceleration value during this time, calculates the motion velocity and trajectory, and matches with the finite set of motion instructions in the pattern matching database to obtain a certain set of motion directions a1 (kxl, Kyl,kzl).

(Axj，Ayj,Azj) = (axj,ayj,azj) - (gx2,gy2,gz2);(Axj, Ayj, Azj) = (axj, ayj, azj) - (gx2, gy2, gz2);

当静止-运动检测模块检测到运动结束时，将结束时的静止姿态作为第二组静止姿态 μ2，并用这个姿态 μ2的三轴加速度传感器数据，作为重力加速度 g2在姿态 μ2时在 x/y/z轴的分量 (gx2,gy2,gz2)，并对运动结束之前的较短固定时间内的三轴加速度数据 (axj,ayj,azj)，去重力加速度 (_gx2,gy2,gz2)，When the stationary-motion detecting module detects the end of the motion, the stationary posture at the end is taken as the second group of stationary postures μ2, and the triaxial acceleration sensor data of the posture μ2 is used as the gravitational acceleration g2 at the attitude μ2 at x/y/ The z-axis component (gx2, gy2, gz2), and the triaxial acceleration data (axj, ayj, azj), and the gravitational acceleration (_g x2, gy2, gz2) for a short fixed time before the end of the motion,

(Axj,Ayj，Azj)得到这段时间内的运动加速度数值、运动速度和轨迹，并且再次与模式匹配数据库当中的有限组运动指令进行匹配，得出另一组确定的运动方向 cc2(kx2，ky2,kz2)。在计算出前后两组运动方向 αΐ 和 α2后，计算 αΐ 到 α2的转换矩阵 Tr，(Axj, Ayj, Azj) obtains the motion acceleration values, motion speeds and trajectories during this time, and again matches the finite set of motion instructions in the pattern matching database to obtain another set of determined motion directions cc2 (kx2, Ky2, kz2). After calculating the motion directions αΐ and α2 of the two groups before and after, the conversion matrix Tr of αΐ to α2 is calculated,

(kx2,ky2,kz2) = (kxl,kyl,kzl) * Tr; (kx2, ky2, kz2) = (kxl,kyl,kzl) * Tr;

在运动开始到结束共有 M个釆样点，将方向的变化矩阵 Tr分配为 M份 Tri， (Tri) = Tr, 计算出每个采样点时刻设备的空间姿态 μί。用 Tri得到的 μί,对完整运动数据中每一点运动加速度进行转化修正，得到最后得到完整运动加速度数据，经过一次积分计算出运动速度，再经过一次积分计算出运动轨迹。。最终实际完整人体运动过程中的加速度、速度、位移，连同设备的空间姿态 μί，共同组成了完整运动数据。There are a total of M sample points from the beginning to the end of the motion, and the direction change matrix Tr is assigned to M parts Tri, (Tri) = Tr, and the spatial attitude μί of the device at each sampling point time is calculated. Using the μί obtained by Tri, the motion acceleration of each point in the complete motion data is transformed and corrected, and the complete motion acceleration data is obtained finally. After one integral, the motion speed is calculated, and then the motion trajectory is calculated after one integral. . Finally, the acceleration, velocity, and displacement during the actual complete human motion, together with the spatial attitude of the device, constitute the complete motion data.

(5) 动作模式匹配模块：(5) Action mode matching module:

参见图 8，其为动作模式匹配模块。动作模式匹配模块利用处理后传入的完整运动数据。根据计算机模式匹配理论，模式匹配模块在操作指令数据库与运动速度轨迹数据库中进行匹配度计算查询并建立映射，得到操作指令数据库中对应的动作操作模式，并向应用程序提供操作指令。针对体感游戏应用的需求，本发明中的解决方案可以将计算出的速度和轨迹信息，通过数据训练和实验得到的先验数据组成匹配数据库，将速度和轨迹映射为游戏场景中的有限组动作操作。例如：体育运动类游戏中的挥臂、抛投、脚步触球等动作模式；第一人称射击游戏中的瞄准、射击等动作模式，即在不改变游戏程序传统的逻辑和流程的前提下对人四肢、头部以及躯干部分的运动进行跟踪识别，从而实现无按键、实时、真实的体感游戏体验。对于拟人类型机器人控制应用的需求，可以通过数据训练和实验得到的先验数据组成匹配数据库，将人体动作映射为机器人操作指令，例如：抬臂、挥臂、降臂等，本发明的解决方案可以大大简化控制设备的体积，增强交互控制的流畅性和仿真性，使得人对于机器人的控制能力和体验大大提高。针对触觉交互应用，本发明可用于采集操作末端姿态信息，将计算出的姿态信息映射到虚拟场景中，与虚拟场景进行动态实时碰撞检测，实现实时触觉交互。对于传统的个人计算机桌面应用场景，本发明可以实现二维平面和三维立体视界内的鼠标控制功能，不但可以完全替代现有传统鼠标的操作功能，而且更适应未来三维立体桌面环境的应用需求。Referring to Figure 8, it is an action pattern matching module. The action pattern matching module utilizes the complete motion data passed in after processing. According to the computer pattern matching theory, the pattern matching module performs a matching degree calculation query and establishes a mapping in the operation instruction database and the motion speed trajectory database, obtains a corresponding action operation mode in the operation instruction database, and provides an operation instruction to the application program. For the needs of the somatosensory game application, the solution in the present invention can compose the calculated speed and trajectory information, the a priori data obtained through data training and experiment into a matching database, and map the speed and trajectory into a limited set of actions in the game scene. operating. For example: sporting games, swinging, throwing, footsteps and other action modes; first-person shooter games, aiming, shooting and other action modes, that is, without changing the logic and flow of the game program tradition The movements of the limbs, head and torso are tracked and identified to achieve a buttonless, real-time, realistic somatosensory gaming experience. For the needs of anthropomorphic robot control applications, a priori data obtained through data training and experiments can be combined to form a matching database, and human motion can be mapped to robot operation instructions, such as: lifting arm, swing arm, descending arm, etc., the solution of the present invention It can greatly simplify the volume of the control device, enhance the fluency and simulation of the interactive control, and greatly improve the control ability and experience of the robot. For haptic interaction applications, the present invention can be used to collect operational end pose information, The calculated posture information is mapped into the virtual scene, and dynamic real-time collision detection is performed with the virtual scene to realize real-time tactile interaction. For the traditional personal computer desktop application scenario, the invention can realize the mouse control function in the two-dimensional plane and the three-dimensional stereoscopic horizon, which can completely replace the operation function of the existing traditional mouse, and is more suitable for the application requirements of the future three-dimensional desktop environment.

(6) 动作显示模块(6) Action display module

参见附图 9，其为动作显示模块。动作显示模块有两个并行的数据处理模块：实时预动作显示模块和完整动作显示模块，对应的输入数据分别来自静止-运动检测模块的预运动数据，以及重力加速度分离模块的完整运动速度、轨迹数据。实时预动作显示模块是将静止运动检测模块进入预运动状态后的前一组静止数据和当前组的运动数据进行简单的去重力加速度处理 (即假设在运动的初始状态，重力加速度在三轴 X/Y/Z的分布不变，即将前面静止状态下 X/Y/Z的加速度数据值作为重力加速度 g在三轴上的分量，将接收到的加速度数据减去重力加速度 g的三轴上的分量)，计算速度、轨迹得到起始运动的方向，并将数据发送到显示驱动以开始绘图。这样设计的目的在于，当用户使用设备开始运动时，初始运动的方向和速度即能够显示出来，无需数据要等到完整动作结束后，经过重力加速度分离模块处理才能显示。这样给用户一种近似实时性的用户体验效果。完整动作显示模块负责将重力加速度分离模块的加速度数据根据采样时间间隔，计算出速度和运动轨迹，将计算得到的数据发送到显示驱动，并在三维动画中绘制和重建用户的运动。可编程系统应用接口（图 1上） Referring to Figure 9, it is an action display module. The motion display module has two parallel data processing modules: a real-time pre-action display module and a complete motion display module, and the corresponding input data are respectively from the pre-motion data of the stationary-motion detection module, and the complete motion speed and trajectory of the gravity acceleration separation module. data. The real-time pre-action display module performs simple de-gravity acceleration processing on the previous set of still data and the current group of motion data after the static motion detecting module enters the pre-motion state (ie, assumes that the initial state of motion, the gravitational acceleration is on the three-axis X The distribution of /Y/Z is constant, that is, the acceleration data value of X/Y/Z in the front stationary state is used as the component of the gravitational acceleration g on the three axes, and the received acceleration data is subtracted from the three axes of the gravitational acceleration g. Component), calculate the speed, the trajectory to get the direction of the starting motion, and send the data to the display driver to start the drawing. The purpose of this design is that when the user starts using the device to start motion, the direction and speed of the initial motion can be displayed, and the data does not need to wait until the completion of the complete motion, and can be displayed after being processed by the gravity acceleration separation module. This gives the user an approximate real-time user experience. The complete motion display module is responsible for calculating the velocity and motion trajectory of the acceleration data of the gravity acceleration separation module according to the sampling time interval, sending the calculated data to the display driver, and drawing and reconstructing the user's motion in the three-dimensional animation. Programmable System Application Interface (Figure 1)

参见图 1,本发明所设计的系统架构和算法解决方案，面对开发者提供了可编程系统应用接口，具备很好的扩展性和应用性。只需针对应用场景，建立模式匹配数据库，并提供硬件方案，则任何有能力的幵发者均可以方便的使用该接口进行编程，用于开发各类应用（体感游戏、机器人遥控器、 3D桌面控制器等），因而本发明具备良好的商业应用前景。结合上述的描述，我们可以得到本发明的有益效果，即可以精确的定位和识别人体的动作，方便的在游戏控制、机器人控制、计算机桌面操作中提供动作识别，该识别系统的数据采集传输模块仅釆用三轴加速度传感器，避免了使用陀螺仪带来的体积和成本的限制，并且在实时性和精度上都达到了很好的效果。本发明中的主要技术在于下面几点 - 1、非陀螺仪去重力加速度算法Referring to Figure 1, the system architecture and algorithm solution designed by the present invention face the developer. Provides a programmable system application interface with excellent scalability and applicability. Just create a pattern matching database for the application scenario, and provide a hardware solution, so any capable sender can easily use this interface for programming, for developing various applications (soso games, robot remote control, 3D desktop) The controller, etc.), thus the present invention has a good commercial application prospect. Combined with the above description, we can obtain the beneficial effects of the present invention, that is, it can accurately locate and recognize the movement of the human body, and conveniently provide motion recognition in game control, robot control, computer desktop operation, and the data acquisition and transmission module of the identification system. The use of a three-axis accelerometer alone avoids the size and cost constraints of using a gyroscope, and achieves good results in both real-time and precision. The main technology in the present invention lies in the following points - 1. Non-gyro de-gravity acceleration algorithm

( 1 )基于三轴加速度传感器的有效数据信号，通过运动转角极小优化算法，求解出一段完整动作从开始到结束过程中，绕某方向转动的最小角度，并利用求解出的轴的矢量方向和转角的大小，将转动分量分离到每一个采样点上，将每一点的姿态和重力加速度建立相应的映射，去除不同姿态下重力加速度对信号的影响，得到准确的运动加速度信息，再计算运动的速度和轨迹。 (1) Based on the effective data signal of the three-axis accelerometer, the minimum angle of rotation of a complete motion from the beginning to the end is calculated by the motion angle mini-optimization algorithm, and the vector direction of the solved axis is utilized. And the size of the corner, the rotation component is separated into each sampling point, the attitude of each point and the gravity acceleration are mapped accordingly, the influence of gravity acceleration on the signal in different postures is removed, the accurate motion acceleration information is obtained, and the motion is calculated. Speed and trajectory.

( 2 ) 基于三轴加速度传感器的有效数据信号，将运动开始之前，结束后的静止数据作为初始和结束的两组静止姿态，并用这个姿态的数据把临近较短固定时间内的加速度数据去除重力加速度，得到这段时间内的运动加速度数值、运动速度和轨迹，并且与模式匹配数据库当中的有限数量的运动指令进行匹配，得出确定的运动方向。在计算出前后两组运动方向后，系统将方向的变化分割并修正完整运动数据中各点速度和轨迹，得到运动信息。对加速度信号进行匹配滤波处理，对连续的数据根据图 4中判断条件 I/II 的对应使用，即判断条件 I触发运动开始状态的检测，判断条件 II触发运动结束状态的检测，避免了人手的小幅震颤以及物体固有震动的干扰，从而识别静止检测和动作起止的算法。下面描述本发明的加速度动作识别系统和方法的具体应用。游戏动作识别设备：(2) Based on the effective data signal of the three-axis acceleration sensor, the static data before the start of the motion is used as the initial and final two sets of static attitudes, and the acceleration data of the shorter fixed time is removed from the gravity using the data of the attitude. Acceleration, the motion acceleration value, motion velocity and trajectory during this time are obtained, and matched with a limited number of motion commands in the pattern matching database to obtain a determined motion direction. After calculating the motion directions of the two groups, the system divides the change of direction and corrects the velocity and trajectory of each point in the complete motion data to obtain motion information. The acceleration signal is matched and filtered, and the continuous data is used according to the corresponding condition of the judgment condition I/II in FIG. 4, that is, the condition I triggers the detection of the motion start state, and the condition II triggers the motion end state detection, thereby avoiding the human hand. A small tremor and interference from the inherent vibration of the object, thereby identifying algorithms for stationary detection and motion start and stop. Specific applications of the acceleration motion recognition system and method of the present invention are described below. Game action recognition device:

本发明可以用于计算机游戏或家庭游戏机游戏中，二维及三维动作的识别。当用户做出模拟真实场景的动作时，具有非陀螺仪技术的加速度动作识别技术的游戏动作识别设备将用户做出的动作通过计算在计算机上重建出来，并与游戏中的控制指令建立映射。对于游戏中二维动作识别，例如扑克牌游戏中对于扑克牌的选取，用户可以将设备佩戴于手背上方，当用户左右平移时，可以对游戏当中手中持有的扑克牌进行选择，当用户希望打出或需选取指定的扑克牌时，用户只需要向前或向后快速移动佩戴有设备的手，即可让计算机识别出相应的操作。这种方式更接近于真实场景中打扑克牌的手部动作。对于三维动作识别，例如打网球的挥拍动作，用户可以将设备佩戴于手背上方，并且可以选择手握网球拍或其他任何物体，当用户根据游戏画面的提示做出一次完整的挥拍动作时，游戏动作识别设备将挥拍过程中的有效数据通过静止 -运动检测模块分离出来，并经过系统计算得到用户挥动球拍的完整轨迹、速度等相关信息。动作模式匹配模块根据运动速度、轨迹信息，计算得到击球的方向、力量以及旋转。具有三维鼠标功能的设备： The present invention can be used for the recognition of two-dimensional and three-dimensional motions in computer games or home game console games. When the user makes an action simulating a real scene, the game motion recognition device having the acceleration motion recognition technology of the non-gyro technology reconstructs the action made by the user on the computer by calculation and maps with the control command in the game. For in-game two-dimensional motion recognition, such as the selection of playing cards in a poker game, the user can wear the device above the back of the hand, and when the user pans left and right, the playing cards held in the hand of the game can be selected, when the user wishes When playing or selecting a designated playing card, the user only needs to quickly move the device-worn hand forward or backward to allow the computer to recognize the corresponding operation. This approach is closer to the hand movements of playing cards in real scenes. For three-dimensional motion recognition, such as a swing action of playing tennis, the user can wear the device above the back of the hand, and can choose to hold the tennis racket or any other object, when the user makes a complete swing action according to the prompt of the game screen. The game motion recognition device separates the valid data in the swing process by the stationary-motion detection module, and obtains the complete trajectory, speed and other related information of the user waving the racket through the system calculation. The motion pattern matching module calculates the direction, strength, and rotation of the shot based on the motion speed and the trajectory information. A device with a three-dimensional mouse function:

传统的鼠标是在二维平面上通过机械或光学解析，得到用户在水平平面上的平移。具有非陀螺仪技术的加速度动作识别技术的鼠标设备，具有在三维空间内精确定位的功能。当用户将持有设备的手向三维空间中某一个方向运动时，鼠标的光标也会随之移动。对于传统的二维操作界面，设备和系统能够将三维运动转化为二维平面的投影，对于鼠标左右键点击的动作，用户可以通过快速甩动手掌或者在垂直于二维平面的方向上下移动，就可以实现相应的点击功能。同时，具有非陀螺仪技术的加速度动作识别技术的鼠标不会像传统的机械或光学鼠标，受到接触表面的限制。具有本发明技术的鼠标可以在任何表面，甚至空中控制计算机桌面的鼠标光标。机器人动作控制输入设备：Conventional mice are mechanically or optically resolved on a two-dimensional plane to give the user a translation in a horizontal plane. Mouse setting with acceleration motion recognition technology with non-gyro technology It has the function of precise positioning in three-dimensional space. When the user moves the hand holding the device to one of the three-dimensional directions, the mouse cursor moves with it. For the traditional two-dimensional operation interface, the device and system can convert the three-dimensional motion into a two-dimensional plane projection. For the left and right mouse click actions, the user can quickly move the palm or move up and down in a direction perpendicular to the two-dimensional plane. You can achieve the corresponding click function. At the same time, the mouse with the acceleration motion recognition technology of non-gyro technology is not limited by the contact surface like a conventional mechanical or optical mouse. A mouse having the technology of the present invention can control the mouse cursor of a computer desktop on any surface, even in the air. Robot motion control input device:

对于拟人类型机器人控制应用的需求，用户可以通过移动佩戴有机器人动作控制输入设备的手臂，将人体动作映射为机器人操作指令，例如：抬臂、挥臂、降臂等。当用户想要对机器人的四肢动作进行控制时，不再需要输入大量复杂的指令，仅仅需要做出相应的动作，就可以控制机器人做出一致的复杂手臂动作。具有本发明技术的机器人动作控制输入设备，相对于 6 维机械传动输入设备等传统解决方案，可以大大简化控制设备的体积，增强交互控制的流畅性和仿真性，使得人对于机器人的控制能力和体验大大提高。对于本领域技术人员显而易见的是，在本发明中可以进行多种修改和变化，而不脱离本发明的精神和范围。因此，本发明意图覆盖落在所附权利要求及其等效范围内的本发明的修改和变化。 For the needs of anthropomorphic robot control applications, the user can map the human motion to robot operation commands by moving the arm with the robot motion control input device, such as: raising the arm, swinging the arm, descending arm, and the like. When the user wants to control the movements of the robot's limbs, it is no longer necessary to input a large number of complicated instructions, and only need to make corresponding actions, the robot can be controlled to make consistent and complex arm movements. The robot motion control input device with the technology of the present invention can greatly simplify the volume of the control device, enhance the fluency and simulation of the interaction control, and improve the control ability of the robot with respect to the conventional solution such as the 6-dimensional mechanical transmission input device. The experience has greatly improved. It will be apparent to those skilled in the art that various modifications and changes can be made in the present invention without departing from the spirit and scope of the invention. Therefore, it is intended that the present invention cover the modifications and modifications of the invention

Claims

权利要求 Rights request

1. 一种动作识别方法，包括下列步骤： 1. A motion recognition method, comprising the following steps:

通过三轴加速度传感器采集一个动作的有效数据信号； Acquiring an effective data signal of an action through a three-axis acceleration sensor;

确定该动作的起始和结束的静止状态数据； Determining the quiescent state data of the start and end of the action;

基于三轴加速度传感器釆集的有效数据信号和该动作的起始和结束的静止状态数据，通过去重力加速度分离算法，将重力加速度从三轴加速度传感器采样的数据中分离出去，得到动作的运动加速度数据；以及 Based on the effective data signal of the triaxial acceleration sensor and the static state data of the start and end of the action, the gravity acceleration is separated from the data sampled by the triaxial acceleration sensor by the de-gravity acceleration separation algorithm to obtain the motion of the action. Acceleration data;

基于该动作的运动加速度数据计算出该动作的速度和轨迹。 The speed and trajectory of the motion are calculated based on the motion acceleration data of the motion.

2. 根据权利要求 1的方法，其中所述去重力加速度分离算法包括：求解出所述动作从开始到结束过程中，所述动作的轴的矢量方向和该轴绕某方向转动的转角的最小角度；2. The method according to claim 1, wherein said de-gravity acceleration separation algorithm comprises: solving a minimum of a vector direction of an axis of said motion and a rotation angle of said axis about a direction from said beginning to the end of said action Angle

利用求解出的轴的矢量方向和转角的角度，将转动分量分离到每一个釆样点上；以及 Separating the rotational component to each of the sample points using the vector direction of the solved axis and the angle of the corner; and

将每一点的姿态和重力加速度建立相应的映射，去除不同姿态下重力加速度对信号的影响，得到准确的运动加速度信息。 Corresponding mapping is established between the attitude of each point and the acceleration of gravity, and the influence of gravity acceleration on the signal in different postures is removed, and accurate motion acceleration information is obtained.

3. 根据权利要求 1的方法，其中所述去重力加速度分离算法包括：基于三轴加速度传感器的有效数据信号，将运动开始之前，结束后的静止状态数据作为初始和结束的两组静止姿态；3. The method according to claim 1, wherein the de-gravity acceleration separation algorithm comprises: based on the effective data signal of the triaxial acceleration sensor, the stationary state data before the start of the motion as the initial and ending two sets of stationary postures;

利用所述两组静止的姿态的数据将临近较短固定时间内的加速度数据去除重力加速度，得到这段时间内的运动加速度数值、运动速度和轨迹； Using the data of the two sets of static attitudes to remove the acceleration data of the acceleration signal for a short fixed time, the motion acceleration value, the motion speed and the trajectory during this period are obtained;

将上述运动加速度数值、运动速度和轨迹与模式匹配数据库当中的有限数量的运动指令进行匹配，得出确定的运动方向；以及 Matching the above-described motion acceleration values, motion speeds, and trajectories with a limited number of motion commands in the pattern matching database to derive a determined direction of motion;

基于计算出的前后两组运动方向，将方向的变化分割并修正完整运动数据中各点速度和轨迹，得到运动信息。Based on the calculated two groups of motion directions, the direction changes are segmented and the velocity and trajectory of each point in the complete motion data are corrected to obtain motion information.

4. 根据权利要求 2或 3的方法，其中所述动作的有效数据信号和动作的起始和结束的静止状态数据具有如下格式：数据的开始和结束分别包括一组静止状态的数据，中间的各组数据是人体运动中三轴加速度传感器采样得到的有效数据。4. A method according to claim 2 or 3, wherein the active data signal of the action and the quiescent state data of the start and end of the action have the following format: the beginning and the end of the data respectively comprise a set of data of the quiescent state, the middle Each group of data is valid data obtained by sampling a three-axis acceleration sensor in human motion.

5. 根据权利要求 2或 3的方法，其中所述确定该动作的起始和结束的静止状态数据的步骤进一步包括静止-运动检测步骤，用于检测动作的预静止和预运动状态。The method according to claim 2 or 3, wherein said step of determining the start and end of the motion state data further comprises a stationary-motion detecting step for detecting a pre-stationary and pre-motion state of the motion.

6. 根据权利要求 5的方法，其中所述静止-运动检测步骤包括：当动作为预静止状态时，将连续的 N帧数据封装成一组；判断该组数据的方差大于预定的阈值 Θ ;6. The method according to claim 5, wherein said stationary-motion detecting step comprises: encapsulating successive N frames of data into a group when the action is in a pre-quiescent state; determining that a variance of the set of data is greater than a predetermined threshold Θ;

如果该组数据饿方差大于预定的阈值 Θ，则使检测状态进入预运动状态，将前一组 N帧数据连同本组预运动状态的数据存储到缓存，并且将这段预运动状态的数据直接发送以计算出该动作的速度和轨迹；当动作为预运动状态时，判断此时的数据方差大于预定的阈值，或者判断数据的均值与标准重力加速度的差值的模是否大于预定的差值 Δ ; If the group of data hungry variance is greater than a predetermined threshold Θ, the detection state is entered into the pre-motion state, the data of the previous group of N frames together with the data of the pre-motion state of the group is stored in the buffer, and the data of the pre-motion state is directly Sending to calculate the speed and trajectory of the action; when the action is a pre-motion state, determining whether the data variance at this time is greater than a predetermined threshold, or determining whether the modulus of the difference between the mean value of the data and the standard gravity acceleration is greater than a predetermined difference Δ ;

当该数据的方差小于预定的阈值 θ，并且数据均值与标准重力加速度的差值的模小于预定的差值 Δ, 则使检测状态进入预静止状态。 When the variance of the data is less than the predetermined threshold θ, and the modulus of the difference between the data mean and the standard gravity acceleration is less than the predetermined difference Δ, the detected state is brought into the pre-rest state.

7. 根据权利要求 6的方法，其中所述静止-运动检测步骤进一步包括- 当检测状态进入预静止状态时，对缓存中数据长度进行最后的判断，检测数据长度是否大于 Τ时间内以频率 F釆样的数据的大小，并且判断数据中是否存在一组数据它的均值与标准重力加速度之差的模大于有效运动幅度 Ω;7. The method according to claim 6, wherein said still-motion detecting step further comprises - when the detected state enters a pre-quiescent state, performing a final determination on the length of the data in the buffer, and detecting whether the data length is greater than the time F by the frequency F The size of the data, and whether there is a set of data in the data, the modulus of the difference between its mean value and the standard gravity acceleration is greater than the effective motion amplitude Ω;

如果这两个条件不都满足，将缓存当中的数据清空； If these two conditions are not met, the data in the cache is cleared;

如果同时满足两个条件，发送数据并通过去重力加速度分离算法将数据中的重力加速度分离。If both conditions are met, the data is sent and the gravitational acceleration in the data is separated by a de-gravity acceleration separation algorithm.

8. 一种动作识别系统，包括. ·8. A motion recognition system, including:

数据采集传输模块,其包括三轴加速度传感器,用于发送三轴加速度传感器采集的一个动作的有效数据信号； a data acquisition transmission module comprising a three-axis acceleration sensor for transmitting an effective data signal of an action acquired by the three-axis acceleration sensor;

驱动程序模块，用于将来自所述数据采集传输模块发送的数据信号缓存，并接收来自数据处理模块的数据，用于显示驱动和人机交互设备的驱动； a driver module, configured to buffer a data signal sent from the data acquisition and transmission module, and receive data from the data processing module, for driving the display driver and the human-machine interaction device;

数据处理模块，其包括重力加速度分离模块，所述数据处理模块用于确定三轴加速度传感器所对应的动作的起始和结束的静止状态数据，并基于三轴加速度传感器采集的有效数据信号和该动作的起始和结束的静止状态数据，通过重力加速度分离模块，将重力加速度从三轴加速度传感器采样的数据中分离出去，得到动作的运动加速度数据，并基于该动作的运动加速度数据计算出该动作的速度和轨迹，并将数据传送到所述驱动程序模块。 a data processing module, comprising: a gravity acceleration separation module, wherein the data processing module is configured to determine the start and end static state data of the action corresponding to the three-axis acceleration sensor, and based on the valid data signal collected by the three-axis acceleration sensor and the The static state data of the start and end of the motion is separated from the data sampled by the triaxial acceleration sensor by the gravity acceleration separation module, and the motion acceleration data of the motion is obtained, and the motion acceleration data is calculated based on the motion acceleration data of the motion The speed and trajectory of the action and transfer the data to the driver module.

9. 根据权利要求 8的动作识别系统，其中所述重力加速度分离模块，9. The motion recognition system according to claim 8, wherein said gravity acceleration separation module,

求解出所述动作从开始到结束过程中，所述动作的轴的矢量方向和动作绕该轴转动的转角的最小角度； Solving a minimum angle of a vector direction of the axis of the motion and a rotation angle of the motion about the axis from the beginning to the end of the action;

10. 根据权利要求 8 的动作识别系统，其中所述重力加速度分离模块，10. The motion recognition system according to claim 8, wherein said gravity acceleration separation module,

基于三轴加速度传感器的有效数据信号，将运动开始之前，结束后的静止状态数据作为初始和结束的两组静止姿态； Based on the effective data signal of the three-axis acceleration sensor, the stationary state data before the start of the motion is taken as the initial and final two sets of stationary postures;

利用所述两组静止的姿态的数据将临近较短固定时间内的加速度数据去除重力加速度，得到这段时间内的运动加速度数值、运动速度和轨迹；Using the data of the two sets of stationary postures to remove the acceleration data of the acceleration data in a short fixed time, and obtain the motion acceleration value and the motion speed during the period. And trajectory;

基于计算出的前后两组运动方向，将方向的变化分割并修正完整运动数据中各点速度和轨迹，得到运动信息。 Based on the calculated two groups of motion directions, the direction changes are segmented and the velocity and trajectory of each point in the complete motion data are corrected to obtain motion information.

1 1. 根据权利要求 9或 10所述的动作识别系统，其中所述动作的有效数据信号和动作的起始和结束的静止状态数据具有如下格式：数据的幵始和结束分别包括一组静止状态的数据，中间的各组数据是人体运动中三轴加速度传感器采样得到的有效数据。1 1. The motion recognition system according to claim 9 or 10, wherein the active data signal of the action and the quiescent state data of the start and end of the action have the following format: the start and end of the data respectively comprise a group of still The data of the state, the data of each group in the middle is the valid data obtained by sampling the three-axis acceleration sensor in human motion.

12. 根据权利要求 9或 10所述的动作识别系统，其中所述数据处理模块进一步包括静止-运动检测模块，用于检测动作的预静止和预运动状态。12. The motion recognition system of claim 9 or 10, wherein the data processing module further comprises a stationary-motion detection module for detecting pre-stationary and pre-motion states of the motion.

13. 根据权利要求 1 1的动作识别系统，其中所述静止 -运动检测模块，13. The motion recognition system of claim 1 wherein said stationary motion detection module,

当动作为预静止状态时，将连续的 N帧数据封装成一组；判断该组数据的方差大于预定的阔值 Θ; When the action is in a pre-quiescent state, the continuous N frame data is encapsulated into a group; determining that the variance of the group of data is greater than a predetermined threshold Θ;

如果该组数据饿方差大于预定的阈值 θ，则使检测状态进入预运动状态，将前一组 Ν帧数据连同本组预运动状态的数据存储到缓存，并且将这段预运动状态的数据直接发送以计算出该动作的速度和轨迹；当动作为预运动状态时，判断此时的数据方差大于预定的阈值，或者判断数据的均值与标准重力加速度的差值的模是否大于预定的差值 Δ; If the group of data is greater than the predetermined threshold θ, the detection state is entered into the pre-motion state, the data of the previous group of frame data is stored in the buffer together with the data of the pre-motion state of the group, and the data of the pre-motion state is directly Sending to calculate the speed and trajectory of the action; when the action is a pre-motion state, determining whether the data variance at this time is greater than a predetermined threshold, or determining whether the modulus of the difference between the mean value of the data and the standard gravity acceleration is greater than a predetermined difference Δ;

当该数据的方差小于预定的阈值 θ，并且数据均值与标准重力加速度的差值的模小于预定的差值 Δ，则使检测状态进入预静止状态。 When the variance of the data is less than the predetermined threshold θ, and the modulus of the difference between the data mean and the standard gravity acceleration is less than the predetermined difference Δ, the detected state is brought into the pre-rest state.

14. 根据权利要求 13的动作识别系统，其中所述静止 -运动检测模块进一步包括，当检测状态进入预静止状态时，对缓存中数据长度进行最后的判断，检测数据长度是否大于 T时间内以频率 F采样的数据的大小，并且判断数据中是否存在一组数据它的均值与标准重力加速度之差的模大于有效运动幅度 Ω;14. The motion recognition system of claim 13 wherein said stationary-motion detection module further comprises When the detection state enters the pre-stationary state, the final judgment is made on the length of the data in the buffer, whether the length of the detected data is greater than the size of the data sampled by the frequency F in the T time, and whether there is a set of data in the data, and the mean and standard of the data. The modulus of the difference in gravitational acceleration is greater than the effective motion amplitude Ω;

如果同时满足两个条件，发送数据并通过去重力加速度分离算法将数据中的重力加速度分离。 If both conditions are met, the data is sent and the gravitational acceleration in the data is separated by a de-gravity acceleration separation algorithm.

15. 根据权利要求 8 的动作识别系统，其中所述数据采集传输模块进一步包括：15. The motion recognition system of claim 8, wherein the data acquisition transmission module further comprises:

微型处理器，其以固定频率 F去读取三轴加速度传感器的数据存储单元，同时根据公共密钥和加密算法对每组数据进行加密，并且将加密过的数据在微型控制器的数据存储单元缓存起来，然后将加密过的数据发送到数据传输模块； a microprocessor that reads the data storage unit of the triaxial acceleration sensor at a fixed frequency F, encrypts each set of data according to a public key and an encryption algorithm, and encrypts the data in a data storage unit of the microcontroller Cache up, and then send the encrypted data to the data transmission module;

数据传输模块，其接收来自微型处理器的数据，采用无线或有线方式发送数据。 A data transmission module that receives data from the microprocessor and transmits the data wirelessly or by wire.

16. 根据权利要求 8 的动作识别系统，其中所述动作识别系统进一步包括：16. The motion recognition system of claim 8 wherein said motion recognition system further comprises:

数据接收模块，用于接收来自数据传输模块的数据，并将其发送到驱动程序模块。 A data receiving module, configured to receive data from the data transmission module and send it to the driver module.

17. 根据权利要求 8 的动作识别系统，其中所述驱动程序模块进一步包括加密 /解密过滤驱动模块，基于微型处理器中的加密算法和公共密钥对所接收的数据进行解密，并对解密后的数据进行校验，当数据量达到触发发送的门限吋，将数据发送至静止-运动检测模块。17. The motion recognition system according to claim 8, wherein said driver module further comprises an encryption/decryption filter driver module, decrypting the received data based on an encryption algorithm and a public key in the microprocessor, and after decrypting The data is verified, and when the amount of data reaches the threshold for triggering transmission, the data is sent to the stationary-motion detection module.

18. 根据权利要求 8 的动作识别系统，其中所述动作识别系统进一步包括：18. The motion recognition system of claim 8 wherein said motion recognition system further comprises:

动作模式匹配模块，用于接收来自重力加速度分离模块的运动加速度数据，计算出动作的速度及空间轨迹，在操作指令数据库与运动速度轨迹数据库中进行匹配度计算査询并建立映射，得到操作指令数据库中对应的动作操作模式数据，并将其发送到驱动程序模块。An action pattern matching module for receiving motion from the gravity acceleration separation module Speed data, calculate the speed and space trajectory of the action, perform the matching degree calculation query in the operation command database and the motion speed trajectory database and establish a mapping, obtain the corresponding action operation mode data in the operation instruction database, and send it to the driver Program module.

19. 根据权利要求 18的动作识别系统，其中所述动作识别系统进一步包括：19. The motion recognition system of claim 18, wherein said motion recognition system further comprises:

动作显示模块，其包括两个并行的数据处理模块，实时预动作显示模块和完整动作显示模块，分别接收来自静止-运动检测模块的预运动数据，以及重力加速度分离模块的完整运动速度、轨迹数据，用于显示动作的运动方向和速度。 The action display module comprises two parallel data processing modules, a real-time pre-action display module and a complete action display module, respectively receiving pre-motion data from the static-motion detection module, and the complete motion speed and trajectory data of the gravity acceleration separation module , used to display the direction and speed of motion of the action.

20. 根据权利要求 19的动作识别系统，其中所述驱动程序模块进一步包括：20. The motion recognition system of claim 19, wherein said driver module further comprises:

人机交互设备驱动模块，用于接收来自动作模式匹配模块的动作操作模式数据，遵循操作系统中标准的输入设备驱动模型要求，提供了面向应用程序的通用控制接口； The human-machine interaction device driving module is configured to receive the action operation mode data from the action mode matching module, and follow a standard input device driver model requirement in the operating system, and provide a general control interface for the application program;

显示驱动模块，用于接收来自所述动作显示模块的动作的运动方向和速度信号，用于驱动在显示屏上动作的显示。 And a display driving module, configured to receive a motion direction and a speed signal from the action display module for driving the display of the action on the display screen.