Method for calibrating plurality of wearable inertial measurement units

Abstract

The invention relates to a method for calibrating a plurality of wearable inertial measurement units, comprising the steps of: (A) aligning the plurality of wearable inertial measurement units with respect to an orientation direction; (B) initializing a coordinate system of the wearable inertial measurement unit; (C) assigning each wearable inertial measurement unit to an active part of the body; (D) mounting each wearable inertial measurement unit to a respective assigned active part of the body; (E) arranging an active part of the body in a defined manner; and (F) performing installation state calibration on each wearable inertial measurement unit.

Description

Technical Field

[0001] The present invention relates to a calibration procedure for inertial measurement units for 3D motion capture using a network of interconnected devices, and sensor fusion of accelerometers and gyroscopes for orientation assessment. Background Art

[0002] Accurately determining the human body posture in 3D space is very important for applications such as the motion capture industry (3D avatars, animation, film production), healthcare (rehabilitation, gait analysis), augmented reality (AR) and virtual reality (VR). The ability to obtain an affordable low-power wireless system will greatly promote the progress of the above fields.

[0003] This task requires reliably determining the orientation of wearable devices in 3D space. This is achieved by using an inertial measurement unit (IMU) as the hardware basis. A sensor fusion algorithm based on a Kalman filter combines accelerometer and gyroscope data to provide relative orientation data in the form of Euler angles or quaternions.

[0004] The advantage of using orientation data is high relative orientation accuracy in 3D space and low power consumption down to 3 mA (2.68 - 3.44 mA, depending on the number of devices). Combining the star architecture of a smart connected sensor (SCS) network (see Figure 1 ) and a low-power Bluetooth (BLE) connection to the end user, the proposed system provides mobile and accurate quantitative data generation and analysis.

[0005] The current disadvantage of mobile motion capture systems is that the mounting orientation of the devices on body parts is restricted, and the orientation information drifts over time or with movement when not using a magnetometer sensor. Therefore, calibration and drift compensation procedures for sensor orientation are required.

[0006] The present invention describes the installation and drift correction procedures for a network of SCS wearable devices in 3D motion tracking. Summary of the Invention

[0007] The present invention relates to a method for calibrating a plurality of wearable inertial measurement units, the method comprising the following steps:

[0008] -(A) Aligning the plurality of wearable inertial measurement units relative to a facing direction;

[0009] -(B) Initializing the coordinate system of the wearable inertial measurement units;

[0010] -(C) Assigning each wearable inertial measurement unit to an active part of the body;

[0011] -(D) Mounting each wearable inertial measurement unit to the respective assigned active part of the body;

[0012] -(E) arranging the movable part of the body in a specified manner;

[0013] -(F) performing installation state calibration on each wearable inertial measurement unit.

[0014] The present invention provides a calibration procedure for installing and compensating for IMU orientation drift in 3D space, which is applicable to providing reliable 3D virtual avatar visualization. The pre-defined calibration postures allow re-mapping the coordinate system for each device to provide one-to-one 3D virtual avatar visualization and data, which can be directly used for quantitative data analysis. Further ideas based on quaternion-based attitude detection and translational motion prediction are revealed.

[0015] The calibration of SCS installation is based on the fact that the orientation direction of the virtual avatar is defined by the startup process, in which all devices are aligned with the selected orientation direction. When the installation calibration of the device is known, the orientation drift of the sensor is corrected. This two-step process allows the 3D virtual avatar application to run continuously for a long time.

[0016] Embodiments of the present invention provide that, in step A, a plurality of wearable inertial measurement units are aligned by placing them in a common receiving unit. Advantageously, the installation device (such as a box) ensures that the IMUs adopt a specified orientation and position relative to the orientation direction.

[0017] Embodiments of the present invention provide that step F will be repeated after a period of time to compensate for the drift of the orientation. The advantage of repeated calibration is that the drift can also be identified and compensated in the future. This means that the frequency of calibration must be reduced.

[0018] Embodiments of the present invention provide that the wearable inertial measurement units establish leaf nodes for wireless communication with a central node. Description of the Drawings

[0019] Figure 1 Shows the leaf-center design scheme of the SCS network of IMU devices.

[0020] Figure 2a and Figure 2b Shows the two-step calibration of the orientation and drift of multiple wearable IMUs.

[0021] Figure 3 Shows a method for calibrating multiple wearable inertial measurement units according to the present invention. Detailed Description of the Invention

[0022] Figure 1Shows the leaf - center design scheme of the SCS network of the IMU device. The leaf node 100 communicates wirelessly with a single center node 200, for example, via low - power Bluetooth (BLE). The center node 200 is connected to a user device 300, such as a smartphone, laptop, etc.

[0023] Figure 2a and Figure 2b Shows a two - step calibration of the orientation and drift of multiple wearable IMUs.

[0024] Figure 2a Shows the initial orientation of multiple aligned SCS boards 10 placed in an alignment box 20. The alignment box 20 determines the orientation direction of the 3D virtual avatar. The local coordinate system of the sensors is shown as x, y, z vectors. The gravitational direction 40 is directly shown on the tabletop, and the alignment box is located on the tabletop.

[0025] Figure 2b Shows the T - position calibration scheme for mounting the SCS board 10 relative to the orientation direction on the human body. Each wearable inertial measurement unit is assigned and mounted on the active part of the body, which in the described case is a human limb.

[0026] Figure 3 Shows a method for calibrating multiple wearable inertial measurement units according to the present invention.

[0027] The method includes the following steps:

[0028] -(A) Aligning multiple wearable inertial measurement units relative to the orientation direction;

[0029] -(B) Initializing the coordinate system of the wearable inertial measurement units;

[0030] -(C) Assigning each wearable inertial measurement unit to the active part of the body;

[0031] -(D) Mounting each wearable inertial measurement unit to the respective assigned active part of the body;

[0032] -(E) Arranging the active part of the body in a prescribed manner;

[0033] -(F) Performing mounting - state calibration on each wearable inertial measurement unit.

[0034] The purpose of the calibration procedure is to compensate for two types of errors:

[0035] The first error type is the mounting error, that is, the change in the orientation of the sensors in the SCS device when placed in the alignment box on the tabletop and when worn on the body.

[0036] The second error type is the sensor drift error, which is the orientation sensor data error that accumulates over time due to the user of the wearable device performing demonstration movements. This error is caused by the integration operation of the fusion algorithm and the noise in the sensor data.

[0037] Set up the SCS board

[0038] Select the direction in which the user performs the demonstration, which represents the facing direction. Before the demonstration and initializing the coordinate system, all SCS devices must be aligned, as Figure 2a shown.

[0039] Install compensation

[0040] The user can assign a body position to the SCS wearable device by mapping the body position to the corresponding MAC-ID of the SCS wearable device in the configuration file.

[0041] Install the device on the corresponding limb according to the settings in the configuration file.

[0042] The user must stand in a T-pose in the facing direction as Figure 2b shown, and then perform installation calibration by recording the current orientation of the device and remapping the coordinate system accordingly.

[0043] Drift compensation

[0044] During the demonstration, orientation drift occurs due to the integration of gyroscope readings and noise in the IMU. The user performs the T-pose again in the facing direction. Then, drift compensation is applied by subtracting the orientation misalignment of the current SCS board relative to the known calibrated pose.

[0045] Formal description

[0046] The orientation of the 3D mesh of the avatar is defined by the current readings of the game rotation vector sensor (GRV), which is in the form of a unit quaternion This value is directly received from the SCS board as the virtual sensor GRV and applied as the rotation of the mesh from the original avatar pose.

[0047] The update rule for the quaternion to be applied as the mesh rotation is as follows (implemented in the update method of the avatar class):

[0048]

[0049] The local rotation of any sub-mesh connected to the parent mesh is calculated as follows:

[0050]

[0051] Here, is the conjugate operation.

[0052] Here, is the calibration quaternion stored as an avatar attribute and updated by the T-pose reset as follows.

[0053] After the device is turned on and aligned in the startup box and mounted on the body in a pre-defined orientation direction, the mounting of the device will be reset as follows:

[0054] The person stands in the T-pose (or any other pre-defined pose in the 3D view), and a set of current quaternion readings are recorded multiplied by the quaternion for calibrating the drift:

[0055] The quaternion for compensating the mounting is calculated as the conjugate of:

[0056] If drift occurs in the GRV sensor readings during the demonstration operation, the T-pose reset for the drift will be performed similarly.

[0057] The person stands in the T-pose, and a set of current quaternion readings are recorded multiplied by the quaternion for calibrating the mounting:

[0058]

[0059] The quaternion for compensating the drift is calculated as the conjugate of:

[0060] Simple pose detection

[0061] Based on the provided hardware and mathematical framework, pose detection can be performed. The normalized quaternion defines the reference pose. Each device is assigned a normalized quaternion

[0062] To detect the pose, the following procedure will be applied:

[0063] 1. Calculate the relative angle between the devices as the angular part of the quaternion difference where,

[0064]

[0065] 2. Create a lookup table for the j poses p = (p1, p2,... pj), where pj = (θ1, θ2,..., θi), i ∈ [1, N] is the number of sensors mounted at pre-defined positions on the human body.

[0066] 3. The score for a given pose pj can be represented as R and describes the similarity between the current position and the pre-recorded position:

[0067]

Claims

1. A method for calibrating multiple wearable inertial measurement units, comprising the following steps: -(A) Align the multiple wearable inertial measurement units relative to the facing direction; -(B) Initialize the coordinate systems of the wearable inertial measurement units; -(C) Assign each wearable inertial measurement unit to an active part of the body; -(D) Mount each wearable inertial measurement unit to the corresponding assigned active part of the body; -(E) Arrange the active parts of the body in a prescribed manner; -(F) Perform mounting state calibration on each wearable inertial measurement unit.

2. The method for calibrating a plurality of wearable inertial measurement units according to claim 1, wherein In step A, the multiple wearable inertial measurement units are aligned by placing them in a common receiving unit.

3. The method for calibrating a plurality of wearable inertial measurement units according to claim 1 or 2, characterized in that, Step F will be repeated after a period of time to compensate for the drift of the orientation.

4. The method for calibrating a plurality of wearable inertial measurement units according to any one of the above claims 1 to 3, characterized in that, The wearable inertial measurement units establish leaf nodes that communicate wirelessly with a central node.

Images