Classifications

• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
  • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
  • G01C21/10—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration
  • G01C21/12—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning
  • G01C21/16—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation
  • G01C21/183—Compensation of inertial measurements, e.g. for temperature effects
  • G01C21/188—Compensation of inertial measurements, e.g. for temperature effects for accumulated errors, e.g. by coupling inertial systems with absolute positioning systems
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
  • G01C19/00—Gyroscopes; Turn-sensitive devices using vibrating masses; Turn-sensitive devices without moving masses; Measuring angular rate using gyroscopic effects
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
  • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
  • G01C21/10—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration
  • G01C21/12—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning
  • G01C21/16—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation
  • G01C21/165—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation combined with non-inertial navigation instruments
  • G01C21/1654—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation combined with non-inertial navigation instruments with electromagnetic compass
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
  • G01C25/00—Manufacturing, calibrating, cleaning, or repairing instruments or devices referred to in the other groups of this subclass
  • G01C25/005—Manufacturing, calibrating, cleaning, or repairing instruments or devices referred to in the other groups of this subclass initial alignment, calibration or starting-up of inertial devices
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01P—MEASURING LINEAR OR ANGULAR SPEED, ACCELERATION, DECELERATION, OR SHOCK; INDICATING PRESENCE, ABSENCE, OR DIRECTION, OF MOVEMENT
  • G01P13/00—Indicating or recording presence, absence, or direction, of movement

Definitions

• the disclosed implementations relate generally to motion sensors and more specifically to a method, system, and device for implementing motions sensors with drift correction, in some implementations, capable of position tracking, more accurate than the Global Positioning System (GPS), and independent of external reference markers, transponders, or satellites.
• GPS Global Positioning System
• Motion tracking detects the precise position and location of an object by recognizing rotation (pitch, yaw, and roll) and translational movements of the object.
• Inertial tracking is a type of motion tracking that uses data from sensors (e.g., accelerometers, gyroscopes, magnetometers, altimeters, and pressure sensors) mounted on an object to measure positional changes of the object. Some of the sensors are inertial sensors that rely on dead reckoning to operate. Dead reckoning is the process of calculating an object’s current location by using a previously determined position, and advancing that position based upon known or estimated accelerations, speeds, or displacements over elapsed time and course.
• dead reckoning techniques are somewhat effective, they are subject to cumulative error called“drift” Because some IMUs estimate relative position/location by integrating acceleration data twice from an accelerometer, even a small error in acceleration results in compounded, increasing, error in relative position/location that accumulates over time.
• some implementations include a tracking device for tracking location and orientation of an object.
• the device comprises one or more sides that form a predetermined shape.
• the device also comprises a plurality of inertial measurement units (IMU) mounted to the one or more sides of the predetermined shape.
• IMU inertial measurement units
• Each IMU is configured to detect movement of the object and generate inertial output data representing non-linear acceleration and/or angular velocity of the object.
• Each IMU includes a first sub sensor and a second sub-sensor.
• Each IMU is positioned at a predetermined distance and orientation relative to each other and a center of mass of the tracking device.
• the device also comprises a controller communicatively coupled to the plurality of IMUs, the controller configured to perform a sequence of steps.
• the sequence of steps comprises receiving first sub-sensor inertial output data and second sub-sensor inertial output data from each of the plurality of IMUs.
• the sequence of steps also comprises for each IMU: generating calibrated inertial output data based on the first sub-sensor inertial output data and the second sub sensor inertial output data; and, cross-correlating the first sub-sensor inertial output data with the second sub-sensor inertial output data to identify and remove anomalies from the first sub-sensor inertial output data with the second sub-sensor inertial output data to generate decomposed inertial output data.
• the sequence of steps also comprises determining the translational and rotational state of the tracking device based on the decomposed inertial output data from each of the IMUs.
• the sequence of steps also comprises synthesizing first sub-sensor inertial output data and second sub-sensor inertial output data to create IMU synthesized or computed data using a synthesizing methodology based on the translational and rotational state of the tracking device.
• the sequence of steps also comprises calculating a current tracking device rectified data output (also referred to herein as“drift-free or“drift- corrected”) based on the synthesized movement of each of the IMUs, a predetermined position of each of the IMUs and a predetermined orientation of each of the IMUs.
• the sequence of steps also comprises calculating a current location and orientation of an object based on a difference between the current object rectified data output, and a previous object drift-free or rectified data output.
• generating calibrated inertial output data includes applying neural network weights to the first sub-sensor inertial output data and the second sub-sensor inertial output data, wherein the neural network weights are adjusted at a learning rate based on the positional state of the tracking device, calculating a discrepancy value representative of a difference between an actual movement of the object and estimated movement of the object, and removing the discrepancy value from the calibrated inertial output data.
• the neural network weights applied to the first sub-sensor inertial output data and the second inertial output data are based on historical inertial output data from each of the first and second sub sensors.
• the decomposed inertial output data corresponding to the first sub-sensor is calibrated based on the second sub-sensor inertial output data by providing feedback to dynamic-calibration neural network of first sub-sensor.
• cross-correlating the first sub-sensor inertial output data with the second sub-sensor inertial output data includes applying pattern recognition to the second sub-sensor inertial output data to generate a decomposed inertial output data representative of the first sub-sensor inertial output data.
• the first sub sensor inertial output data and second sub-sensor inertial output data are filtered to minimize signal noise through signal conditioning.
• the first sub sensor inertial output data and second sub-sensor inertial output data from each of the plurality of IMUs is received periodically at less than approximately 1 millisecond (ms) for continuous high sampling rate.
• the first sub sensor and the second sub-sensor are each one of: accelerometer, magnetometer, gyroscope, altimeter, and pressure sensor; wherein the first sub-sensor is a different sensor type than the second sub-sensor.
• the predetermined shape is one of: a plane, a tetrahedron, a cube, or any platonic solid, or any other irregular configurations with known distances and angels between IMUs.
• At least some of the IMUs used to calculate the rectified IMU data output are oriented at different angles along two different axes relative to each other.
• calculating the current position and orientation of the object based on the difference between the current rectified IMU output and the previous object rectified IMU output include: identifying an edge condition; and blending the current object rectified IMU output and the previous object rectified IMU output to remove the edge condition using neural networks
• some implementations include a method of tracking the location and orientation of an object using a tracking device.
• the tracking device includes one or more sides that define a predetermined shape.
• the tracking device also includes a plurality of inertial measurement units (IMU) mounted to the one or more sides of the predetermined shape.
• IMU inertial measurement units
• Each IMU includes a first sub-sensor and a second sub-sensor.
• the tracking device also includes a controller communicatively coupled to the plurality of IMUs.
• the method comprises performing a sequence of steps.
• the sequence of steps includes, at each IMU, detecting movement of the object and generating inertial output data representing acceleration and/or angular velocity of the object.
• the sequence of steps also includes, at the controller, receiving first sub-sensor inertial output data and second sub-sensor inertial output data from each of the plurality of
• the sequence of steps also includes, at the controller, for each IMU: generating calibrated inertial output data based on the first sub-sensor inertial output data and the second sub-sensor inertial output data; cross-correlating the first sub-sensor inertial output data with the second sub-sensor inertial output data to identify and remove anomalies from the first sub-sensor inertial output data with the second sub-sensor inertial output data to generate decomposed inertial output data.
• the sequence of steps also includes, at the controller, determining a translational and rotational state of the tracking device based on the decomposed inertial output data from each of the IMUs.
• the sequence of steps also includes, at the controller, synthesizing first sub-sensor inertial output data and second sub-sensor inertial output data to create IMU synthesized or computed data using a synthesizing methodology based on the positional and rotational state of the tracking device.
• the sequence of steps also includes, at the controller, calculating a current tracking device overall drift-free or rectified data output based on the synthesized movement of each of the IMUs, a predetermined location of each of the IMUs and a predetermined orientation of each of the IMUs.
• the sequence of steps also includes, at the controller, calculating a current location and orientation of an object based on a difference between the current object overall rectified data and a previous object overall rectified data.
• generating calibrated inertial output data includes applying neural network weights to the first sub-sensor inertial output data and the second sub-sensor inertial output data, wherein the neural network weights are adjusted at a learning rate based on the positional state of the tracking device, calculating a discrepancy value representative of a difference between an actual movement of the object and estimated movement of the object, and removing the discrepancy value from the calibrated inertial output data.
• the neural network weights applied to the first sub-sensor inertial output data and the second inertial output data are based on historical inertial output data from each of the first and second sub-sensors.
• the decomposed inertial output data corresponding to the first sub-sensor is calibrated based on the second sub-sensor inertial output data by providing feedback to dynamic-calibration neural network of first sub-sensor.
• cross-correlating the first sub-sensor inertial output data with the second sub-sensor inertial output data includes applying pattern recognition to the second sub-sensor inertial output data to generate a decomposed inertial output data representative of the first sub-sensor inertial output data.
• the first sub-sensor inertial output data and the second sub-sensor inertial output data are filtered to minimize signal noise through signal conditioning.
• the first sub-sensor inertial output data and second sub-sensor inertial output data from each of the plurality of IMUs is received periodically at less than approximately 1 ms for continuous high sampling rate.
• the first sub-sensor and the second sub-sensor are each one of: accelerometer, magnetometer, gyroscope, altimeter, and pressure sensor and the first sub-sensor is a different sensor type than the second sub-sensor.
• the predetermined shape is one of: a plane, a tetrahedron, a cube or any platonic solid, or any other irregular configurations with known distances and angels between IMUs.
• IMUs used to calculate the overall drift-free or rectified system output are oriented at different angles along two different axes relative to each other.
• calculating the current location and orientation of the object based on the difference between the current object rectified data and the previous object rectified data output include: identifying an edge condition; and blending the current object rectified data output and the previous object rectified data output to remove the edge condition using neural networks.
• Figures 1A-1F illustrate various configurations of motion sensors mounted on two-dimensional (“2-D”) or three-dimensional (“3-D”) objects, in accordance with some implementations.
• Figure 2 is a block diagram illustrating a representative system with sensor(s) with drift correction, according to some implementations.
• Figure 3 is a flow diagram illustrating the flow of sensor data through a representative system with drift correction, according to some implementations.
• Figures 4A-4D illustrate a flowchart representation of a method of tracking position and orientation of an object using a tracking device, according to some
• first, second, etc. are, in some instances, used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another.
• a first electronic device could be termed a second electronic device, and, similarly, a second electronic device could be termed a first electronic device, without departing from the scope of the various described implementations.
• the first electronic device and the second electronic device are both electronic devices, but they are not necessarily the same electronic device.
• the phrase“if it is determined” or “if [a stated condition or event] is detected” is, optionally, construed to mean“upon determining” or“in response to determining” or“upon detecting [the stated condition or event]” or“in response to detecting [the stated condition or event]” or“in accordance with a determination that [a stated condition or event] is detected,” depending on the context.
• motion sensors that correct for drift
• applications for motion sensors that correct for drift including, but not limited to, gaming systems, smartphones, helmet-mounted displays, military applications, and gesture tracking devices, among others.
• U.S. Patent No. 9,417,693 the“’693 Patent”
• HMI wearable wireless human-machine interface
• a user can control a controllable device based on gestures performed by the user using the wearable HMI.
• a controller to track motion and correct for drift may be connected to the IMUs of the wearable HMI.
• the controller is attached to or integrated in the wearable HMI.
• the controller is remote from the wearable HMI but communicatively coupled to the wearable HMI.
• Figures 1A-1F illustrate various configurations of motion sensors mounted on
• Motion sensors may be mounted in linear arrays, on planar surfaces, or vertices of a myriad of geometric configurations, formed by any dimensional planar surface, platonic solid, or irregular 3D object.
• drift can be eliminated by, among certain methods or portions thereof described herein, resetting the motion sensors’ instantaneous measured acceleration, angular velocity, magnetic orientation, and altitude to match the known geometry formed by the physical distances and angles of the motion sensors relative to each other, as further described below in reference to flowcharts 4A-4D below.
• two sensors 102, 104 are positioned adjacent to each other at a fixed distance 107, and the angles between the two sensors can be considered to be approximately 0 degrees or approximately 180 degrees.
• this drift can be removed and positions of the two motion sensors can be reset to a fairly accurate degree.
• a planar configuration of three (3) or four (4) or more sensors can provide a spatial calculation based on a higher number of IMU readings of instantaneous measurements of all sensors in the array with known physical angles and distances between them.
• Figure 1B shows a four-sensor configuration with sensors 106, 108, 110, and 112 mounted adjacent to each other in a planar configuration.
• Planar configurations such as the configurations shown in Figures 1A and 1B, provide a simpler mathematical model with fairly low demand for computation.
• variations in axial motion detection methods of the physical sensors may affect the accuracy of measurement in different axes of motion and orientation.
• motion in the Z-axis of a MEMS-based sensor is heavily biased with a gravity vector which may introduce higher variance in the physical motion of the sensor in this axis.
• the coriolis force, used to calculate Yaw in the z-axis is also susceptible to larger variance than the X, or Y axis.
• a tetrahedron configuration with four (4) sensors, each one mounted on each face of the tetrahedron can provide a blend of multi-axial data resulting in better complementary and compensatory measurement for the gravity vector bias than a single, Z-Axis of all sensors, according to some implementations.
• Figures 1C and 1D show one such configuration.
• Figure 1C shows the top - oblique view of a tetrahedron with motion sensors 114, 116, 118 mounted on each of the visible three faces.
• Figure 1D shows the bottom - oblique view of the tetrahedron shown in Figure 1C showing the additional sensor 120 on the fourth face of the tetrahedron.
• a component of the X and Y axis is also exposed to the gravity vector from at least three sensors at any given time, permitting a higher degree of accuracy through the removal of the gravity vector from a number of sensors and a number of axes at any instantaneous measurement.
• Sensors are mounted at angles on each surface, providing a blend of X, Y, and Z axis data for better spatial calculations and drift correction, in accordance with some implementations.
• a cubic configuration will provide a higher sensor count of six (6) sensors on the six (6) surfaces of the cube to stabilize the spatial deviation even further.
• Figure 1E shows an oblique view of a cubic configuration, according to some
• each of the six faces may have a sensor mounted, including the sensors 122, 124, and 126.
• some, less than all, faces of any object described herein have at least one sensor.
• each sensor on each face enables a complementary reading between the other sensors on the other faces of the cube.
• the latency to read all measurements is also increased in the cubic or higher dimensional solid geometries.
• Motion sensors can also be rotated on opposite faces of the geometric solids to provide an axial blend in any configuration, according to some implementations.
• Figure 1F shows an oblique view of another configuration of the cuboid in Figure 1E wherein motion sensors are mounted on each face of the cube as before, but sensors may be rotated at an angle between zero (0) and ninety (90) degrees, non-inclusive.
• sensor 122 may be rotated at an angle of approximately forty-five (45) degrees with respect to the other sensors.
• this method may provide a better analysis of instantaneous motion data, the computation time per measurement-to-calculation output may be longer.
• FIG. 2 is a block diagram illustrating a representative system 200 with drift- free sensor(s), according to some implementations.
• the system 200 includes one or more processing units (e.g., CPUs, ASICs, FPGAs, microprocessors, and the like) 202, one or more communication interfaces 214, memory 220, and one or more communication buses 216 for interconnecting these components (sometimes called a chipset).
• the type of processing units 202 is chosen to match the requirement of application, including power requirements, according to some implementations. For example, the speed of the CPU should be sufficient to match application throughput.
• the system 200 includes a user interface 208.
• the user interface 208 includes one or more output devices 210 that enable presentation of media content, including one or more speakers and/or one or more visual displays.
• user interface 208 also includes one or more input devices 212, including user interface components that facilitate user input such as a keyboard, a mouse, a voice-command input unit or microphone, a touch screen display, a touch- sensitive input pad, a gesture capturing device, or other input buttons or controls.
• some systems use a microphone and voice recognition or a camera and gesture recognition or a motion device and gesture recognition to supplement or replace the keyboard.
• the system 200 includes one or more Inertial
• the IMUs include one or more accelerometers, one or more magnetometers, and/or one or more gyroscopes, and/or altimeters, and/or pressure sensors.
• the one or more IMUs are mounted on an object that incorporates the system 200 according to a predetermined shape.
• Figures 1A-1F described above illustrate various exemplary configurations of motion sensors.
• the initial configuration of the IMUs e.g., the number of IMUs, the predetermined shape
• the initial configuration of the IMUs is also determined based on characteristics of the individual IMUs. For example, the orientation or the axis of the IMUs, and therefore the predetermined shape, are chosen so as to compensate for manufacturing defects.
• the one or more IMUs are fabricated as CMOS and MEMS system on a chip (SOC) that incorporates the system 200.
• Communication interfaces 214 include, for example, hardware capable of data communications using any of a variety of custom or standard wireless protocols (e.g., IEEE 802.15.4, Wi-Fi, ZigBee, 6L0WPAN, Thread, Z-Wave, Bluetooth Smart, ISAlOO.l la, WirelessHART, MiWi, etc.) and/or any of a variety of custom or standard wired protocols (e.g., Ethernet, HomePlug, etc.), or any other suitable communication protocol, including communication protocols not yet developed as of the filing date of this document.
• custom or standard wireless protocols e.g., IEEE 802.15.4, Wi-Fi, ZigBee, 6L0WPAN, Thread, Z-Wave, Bluetooth Smart, ISAlOO.l la, WirelessHART, MiWi, etc.
• any of a variety of custom or standard wired protocols e.g., Ethernet, HomePlug, etc.
• Memory 220 includes high-speed random access memory, such as DRAM,
• SRAM static random access solid state memory
• DDR RAM dynamic random access solid state memory devices
• non-volatile memory such as one or more magnetic disk storage devices, one or more optical disk storage devices, one or more flash memory devices, one or more EPROMs, one or more EEPROMs, or one or more other non-volatile solid state storage devices.
• Memory 220 includes a non- transitory computer readable storage medium.
• memory 220 or the non-transitory computer readable storage medium of memory 220, stores the following programs, modules, and data structures, or a subset or superset thereof:
• operating logic 222 including procedures for handling various basic system services and for performing hardware dependent tasks
• device communication module 224 for connecting to and communicating with other network devices (e.g., network interface, such as a router that provides Internet connectivity, networked storage devices, network routing devices, server system, etc.) connected to one or more networks via one or more communication interfaces 214 (wired or wireless); • input processing module 226 for detecting one or more user inputs or interactions from the one or more input devices 212 and interpreting the detected inputs or interactions;
• network devices e.g., network interface, such as a router that provides Internet connectivity, networked storage devices, network routing devices, server system, etc.
• input processing module 226 for detecting one or more user inputs or interactions from the one or more input devices 212 and interpreting the detected inputs or interactions;
• user interface module 228 for providing and displaying a user interface in which settings, captured data, and/or other data for one or more devices (not shown) can be configured and/or viewed;
• one or more application modules 230 for execution by the system 200 for controlling devices, and for reviewing data captured by devices (e.g., device status and settings, captured data, or other information regarding the system 200 and/or other client/electronic devices);
• controller modules 240 which provides functionalities for processing data from the one or more IMUs 204, including but not limited to:
• o data receiving module 242 for receiving data from the one or more IMUs 204 that is to be processed by the controller module(s) 240;
• o filtering module 244 for removing noise from the raw data received by the data receiving module 242;
• o dynamic calibration module 246 for cross-correlating the data between the one or more IMUs 204 (e.g., different gyroscopes and accelerometers of the one or more IMUs 204) to calibrate filtered data for the one or more IMUs 204; o motion decomposition module 248 that determines positional and rotational state based on the decomposed output for each of the one or more IMUs; o motion synthesis module 250 for synthesizing motion based on the output of the dynamic calibration module 246 and the motion decomposition module 248;
• o drift correction module 252 for correcting drift in the sensor output (e.g., using an Adaptive Continuous Fuzzy Rule (without modus ponens) Bayesian Filter with Trapazoidal Motion Paramters (ACFBT) for the predetermined shape based on the output from the motion synthesis module 248; and o edge condition handling module 254 that handles complex movements based on the output (e.g., using Artificial Intelligence/Neural Networks/Deep Learning) of the drift correction module 252.
• ACFBT Adaptive Continuous Fuzzy Rule
• ACFBT Trapazoidal Motion Paramters
• the 242 from the IMUs include acceleration information from accelerometers, angular velocities from gyroscopes, degrees of rotation of magnetic field from magnetometers, atmospheric pressure from Altimeters, and differential Pressure Sensors.
• the raw data is received from each of the IMUs sequentially, according to some implementations.
• the IMU data is received in parallel.
• the filtering module 244 filters the raw data to remove noise from the raw data signals received by the data receiving module 242.
• the filtering module 244 uses standard signal processing techniques (e.g., low-pass filtering, clipping, etc.) to filter the raw data thereby minimizing noise in sensor data, according to some implementations.
• the filtering module 244 also computes moving averages and moving variances using historical data from the sensors, according to some implementations.
• the dynamic calibration module 246 uses an
• AI framework e.g., a neural network framework
• one or more“neurons” are configured in a neural network configuration to calibrate the filtered data for the one or more IMUs 204.
• shape of the object (sometimes herein called a predetermined shape) is a cuboid for the sake of explanation.
• a cuboid shaped object could be placed on a planar surface six different ways (i.e., on six different faces of the cuboid). So there are six orientations to calibrate on.
• the system 200 collects a large number of samples (e.g., approximately 1,000 or more samples) for each of the six orientations. This sampled data is collected and stored in memory 220. Later, when raw data is received, the stored sampled data is used as a baseline to correct any offset error in the raw data during sedentary states (i.e., when the object is not moving).
• the weights of the network are constantly tuned or adjusted based on the received raw data from the IMUs after offsetting the stored sampled data, according to some implementations.
• a neural network-based solution provides better estimates than a least squares regression analysis or statistical measures.
• the neural network weights are adjusted dynamically, consider when the object is stationary but the neural network output indicates that the object is moving. The weights are readjusted, through back propagation, such that the output will indicate that the object is stationary. Thus the weights settle during times when the object is stationary.
• the learning rate of the neural network is maximized during sedentary states (sometimes herein called stationary states), and minimized when the object is in motion. Pattern recognition is used to detect whether the object is moving or is stationary so that the learning rate can be adjusted, according to some implementations.
• the different stationary and mobile states are used to adjust the weights affecting the
• known reference to the magnetic north is used to constantly adjust the weights that correspond to the magnetometers.
• the magnetometer data is also used to correct or settle the weights for the accelerometers when the object is moving because the reference point for the magnetic north and gravity vector are always known.
• Gyroscope data is more reliable than data from accelerometers because it only requires a single level integration. So the gyroscope data is also used to correct accelerometer weights, according to some implementations.
• the dynamic calibration module 246 is optional, and a pass through channel passes the output of the filtering module 244 to the motion synthesis module 250 without dynamic calibration.
• the motion decomposition module 248 uses pattern recognition techniques to eliminate anomalies due to cross-interaction or interference between the sub-sensors, in each IMU.
• Experimental data is collected for controlled translational and rotational movements of an object. For example, the behavior of the gyroscope is tracked under constant velocity and the pattern is stored in memory. When the gyroscopic data follows the known pattern, the fact that the object is under constant velocity is deduced based on this pattern.
• accelerometer data e.g., constant gravity vector
• magnetometer data can be used to identify patterns to correct errors in accelerometer data and/or gyroscope data, according to some implementations.
• the motion decomposition module 248 removes anomalies by observing changes in patterns detected from sensor data, such as when the object stops moving or rotating abruptly, as another effect to correct for anomalies. In some implementations, the motion decomposition module 248 analyzes several distinct stored patterns for correcting anomalies in each of the sensors. In some implementations, the motion decomposition module 248 categorizes the type of translational and/or rotational movements of each IMU of the tracked object and outputs the pattern or the category for the motion synthesis module 250. For example, the motion decomposition module 248 deduces that each IMU is in one of many states, including simple linear motion, simple linear motion with rotation, non-linear motion with simple rotation. In some implementations, output from the motion decomposition module 248 additionally controls the learning rate in the dynamic calibration module 246.
• the motion synthesis module 250 uses the state information (e.g., constant velocity, constant acceleration, changing acceleration, in combination with rotation) from the motion decomposition module 248 to select one or more algorithms/methodologies. The motion synthesis module 250 subsequently applies the one or more algorithms on the data output from dynamic calibration module 246 to synthesize the motion of the object (sometimes herein referred to as the computation of overall rectified data for the one or more IMUs).
• state information e.g., constant velocity, constant acceleration, changing acceleration, in combination with rotation
• the motion synthesis module 250 subsequently applies the one or more algorithms on the data output from dynamic calibration module 246 to synthesize the motion of the object (sometimes herein referred to as the computation of overall rectified data for the one or more IMUs).
• the motion synthesis module 250 uses an equation to compute the axis of rotation based on the difference in angular momentum of the IMUs (as indicated by the output of the dynamic calibration module) and the known shape outlined by the predetermined position of the different IMUs.
• the object is mounted with IMUs in a planar configuration, such as in Figure 1B, with four sensors, each sensor in a comer.
• the planar configuration positioned vertically in a diamond shape, with the longitudinal axis passing through the top IMU and the bottom IMU.
• the side IMUs on either side of the longitudinal axis will share the same angular momentums but will have different angular momentums as compared to the top IMU and the bottom IMU, and the top IMU will have an angular velocity greater than the bottom IMU that is closer to the axis of rotation.
• the motion synthesis module 250 computes or synthesizes the rotational axis data from the differences in the angular momentums and the known distances between the sensors, based on the shape formed by the IMUs.
• the drift correction module 252 uses Bayes theorem to remove drift by re-conforming sensor positions and orientation to the known (sometimes herein called predetermined) shape.
• the Bayesian filter predicts how much the IMU data is drifting by.
• the drift correction module 252 computes the skewness in the data by the motion sensors based on the variation in the norms, distances and angles between the sensors. If the variation in the norms exceeds a threshold, the drift correction module 252 generates a correction matrix (sometimes called a drift matrix) to eliminate drift in successive sensor readings.
• a shape correcting module (not shown) corrects the data output from the dynamic calibration module (sometimes herein called the clean or filtered data) using the correction matrix, by subtracting the predicted drift from the clean data, in a continuous or iterative fashion, according to some implementations. For example, after every reading of sensor data, previously generated and stored data from the drift correction module 252 is used to correct the clean data output from the noise-filtered, and dynamic calibration module, according to some implementations.
• the edge condition handling module 254 handles complex movements (e.g., while spinning along two axes, and moving across on a straight line, say the object also lifts up) and/or transitional movements (e.g., spinning to laterally moving along a straight line) to reduce drift based on the output of the drift correction module 252.
• the edge condition handling module 254 uses AI to apply probability weightings to compensate for the edge conditions.
• the edge condition handling module 254 blends a current object common data point (e.g., output by the drift correction module 252) and the previous object common data point (e.g., previous output for a prior sensor reading by the drift correction module 252 that is stored in memory) to remove the edge condition.
• drift Even though in some complex scenarios, drift may not be completely corrected, perpetual drift or constant drift can be eliminated with the aforementioned modules. Furthermore, in some implementations, the drift observed by the combination of the modules described herein is in the order of centimeters or even millimeters, whereas alternate external reference based drift elimination (e.g., using a GPS) could sometimes result in a drift in the order of meters.
• alternate external reference based drift elimination e.g., using a GPS
• Each of the above identified elements may be stored in one or more of the previously mentioned memory devices, and corresponds to a set of instructions for performing a function described above.
• the above identified modules or programs i.e., sets of instructions
• memory 220 optionally, stores a subset of the modules and data structures identified above.
• memory 220 optionally, stores additional modules and data structures not described above.
• one or more processing modules and associated data stored in the memory 220 are stored in and executed on a second processing device other than the system with drift-free motion sensors 200 that is configured to receive and process signals produced by the IMUs 214.
• the second processing device might be a computer system, smart home device or gaming console that executes applications (e.g., computer games) at least some of whose operations are responsive to motion signals provided by the IMUs.
• FIG. 3 is a flow diagram illustrating the flow of sensor data through a representative system with drift-free sensor(s), according to some implementations.
• Raw data (302) from the one or more IMU sensors (IMU 0, IMU 1, IMU 2, ... , IMU N) is received (324) by the controller 300 (e.g., controller module 240).
• the controller receives the data from the one or more IMUs in parallel (as shown in Figure 3).
• the received data is output as raw data (304) to the motion decomposition module 326, according to some implementations.
• the raw data is also input as data 306 to a filtering module 328 which filters the raw data to produce filtered data 310 which is in turn input to a dynamic calibration module 330.
• the motion decomposition module 326 also controls (314) the learning rate of the dynamic calibration module 330.
• the motion decomposition module 326 and/or the dynamic calibration module 330 are optional modules. In such cases, the filtered data 310 is input (not shown) to the motion synthesis module.
• the motion synthesis module 332 in these cases, does not know the pattern or category of motion but iteratively applies one or more algorithms or equations to synthesize motion.
• the motion decomposition 326 and the dynamic calibration 330 steps execute asynchronously and/or in parallel.
• the Bayes calculation step 336 uses the output 316 of the motion synthesis module to generate drift correction matrices 320 (as described previously with reference to Figure 2) which is consumed by a shape correction module 334 to correct input in the next iteration (i.e., when and after such data becomes available) of motion synthesis.
• the shape correction data is not available, and the dynamic calibration output 312 is input to motion synthesis step 332.
• the output of the Bayes calculation step 336 (318) is input to a step 338 to handle edge conditions (described above in reference to Figure 2) for complex movements and dynamic learning.
• the output 322 indicates drift-free real motion output of the controller, according to some implementations.
• filtering module 328 includes similar functionality to filtering module 244 in Figure 2; motion decomposition module 326 includes similar functionality to motion decomposition module 248 in Figure 2; dynamic calibration module 330 includes similar functionality to dynamic calibration module 246 in Figure 2; shape correction module 334 includes similar functionality to shape correction module described above in the description for Figure 2; motion synthesis module 332 includes similar functionality to motion synthesis module 250 in Figure 2; Bayes calculations module 336 includes similar functionality to drift correction module 252 in Figure 2; and handle edge conditions module 338 includes similar functionality to edge condition handling module 254 in Figure 2.
• FIGs 4A-4D illustrate a flowchart representation of a method 400 of tracking position and orientation of an object using a tracking device, according to some implementations.
• the tracking device includes (402) one or more sides that define a predetermined shape, and a plurality of inertial measurement units (IMU) mounted to the one or more sides of the predetermined shape.
• each IMU includes a first sub-sensor and a second sub-sensor, and each IMU is positioned at a predetermined distance and orientation relative to a center of mass of the tracking system, according to some implementations.
• Figures 1A-1F described above illustrate various configurations of sensors mounted on 3D objects, according to some implementations.
• the first sub-sensor and the second sub-sensor of the tracking device are each one of: accelerometer, magnetometer, gyroscope, altimeter, and pressure sensor and the first sub-sensor is a different sensor type than the second sub-sensor.
• the predetermined shape of the tracking device is one of: a plane, a tetrahedron, and a cube.
• the tracking device also includes a controller communicatively coupled to the plurality of IMUs. An example system 200 with IMUs 204 was described above in reference to Figure 2, according to some implementations.
• each IMU of the tracking device detects movement of the object and generates inertial output data representing location and/or orientation of the object.
• IMUs 204 in Figure 2 or the sensors in Figures 1A-1F use a combination of accelerometers, magnetometers, gyroscopes, altimeters, and/or pressure sensors to detect movement of the object and generate data that represents location and/or orientation of the object.
• the tracking object at the controller (410), receives
• first sub-sensor inertial output data and second sub-sensor inertial output data from each of the plurality of IMUs receives the output from the one or more IMUs 204 via the communication bus 216.
• the controller receives (414) the first sub-sensor inertial output data and the second sub-sensor inertial output data from each of the plurality of IMUs periodically at less than approximately 1 ms for continuous high sampling rate.
• the controller uses a filtering module (e.g., module
• the controller performs a sequence of steps 418 for each IMU, according to some implementations.
• the controller generates (420) calibrated inertial output data based on the first sub-sensor inertial output data and the second sub-sensor inertial output data.
• the controller uses the dynamic calibration module 246 to generate calibrated inertial output data.
• the controller calculates the error value using (422) neural network weights to evaluate the first sub-sensor inertial output data and the second sub-sensor inertial output data, wherein the weights are adjusted at a learning rate based on the positional state (e.g., stationary position state) of the tracking device, calculating a discrepancy value
• the controller applies (424) neural network weights to the first sub sensor inertial output data and the second inertial output data based on historical (e.g., prior or previous) inertial output data from each of the first and second sub-sensors.
• the controller stores and/or accumulates inertial output data received from the IMUs over time that is later retrieved as historical data.
• the controller uses the dynamic calibration module (e.g., module 246) to cross-correlate (426) the first sub-sensor inertial output data with the second sub-sensor inertial output data to identify and remove anomalies from the first sub-sensor inertial output data with the second sub-sensor inertial output data to generate decomposed inertial output data for each IMU, according to some implementations.
• the controller calibrates (428) the decomposed inertial output data corresponding to the first sub-sensor based on the second sub-sensor inertial output data.
• the controller cross-correlates the first sub-sensor inertial output data with the second sub-sensor inertial output data by applying (430) pattern recognition (e.g., by using a motion decomposition module, such as module 248) to the second sub-sensor inertial output data to generate the decomposed inertial output data representative of the first sub sensor inertial output data.
• pattern recognition e.g., by using a motion decomposition module, such as module 248
• the controller determines (432), using a motion decomposition module (e.g., module 248 described above), a positional and rotational state of the tracking device based on the decomposed inertial output data from each of the IMUs, according to some implementations.
• a motion decomposition module e.g., module 248 described above
• the controller synthesizes (434), using a motion synthesis module (e.g., module 250 described above), first sub-sensor inertial output data and second sub-sensor inertial output data to create IMU synthesized data using a synthesizing methodology based on the positional and rotational state of the tracking device, according to some implementations.
• a motion synthesis module e.g., module 250 described above
• first sub-sensor inertial output data and second sub-sensor inertial output data to create IMU synthesized data using a synthesizing methodology based on the positional and rotational state of the tracking device, according to some implementations.
• the controller calculates (436), using a ACFBT calculation module (not shown), a current tracking device rectified data output based on the data synthesized for each of the IMUs, a predetermined position of each of the IMUs and a predetermined orientation of each of the IMUs to confirm to a predetermined shape.
• a current tracking device rectified data output based on the data synthesized for each of the IMUs
• a predetermined position of each of the IMUs and a predetermined orientation of each of the IMUs to confirm to a predetermined shape.
• at least some of the IMUs used to calculate the common data point are oriented at different angles along two different axes relative to each other.
• the controller subsequently calculates (440), using a current position and orientation determination module (e.g., module 252 in Figure 2, or steps 336 and 334 in Figure 3), a current position and orientation of an object based on a difference between the current object rectified data output and a previous object rectified data output, according to some implementations.
• the controller identifies (442) an edge condition (e.g., complex movements described above) and blends (444), using an edge condition handling module (e.g., module 254 described above), the current object rectified data output and the previous object rectified data output to remove the edge condition.
• an edge condition e.g., complex movements described above
• an edge condition handling module e.g., module 254 described above
• FIGS. 4A-4D have been described is merely an example and is not intended to indicate that the described order is the only order in which the operations could be performed.
• One of ordinary skill in the art would recognize various ways to reorder the operations described herein.
• stages that are not order dependent may be reordered and other stages may be combined or broken out. While some reordering or other groupings are specifically mentioned, others will be obvious to those of ordinary skill in the art, so the ordering and groupings presented herein are not an exhaustive list of alternatives.
• the stages could be implemented in hardware, firmware, software or any combination thereof.

Landscapes

• Engineering & Computer Science (AREA)
• Radar, Positioning & Navigation (AREA)
• Remote Sensing (AREA)
• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Automation & Control Theory (AREA)
• Manufacturing & Machinery (AREA)
• Electromagnetism (AREA)
• Navigation (AREA)
• Position Fixing By Use Of Radio Waves (AREA)

Priority Applications (4)

Applications Claiming Priority (2)

Publications (1)

Family

ID=68985327

Family Applications (1)

Country Status (5)

Families Citing this family (10)

Citations (4)

Family Cites Families (27)

• 2019
  • 2019-06-26 US US16/453,961 patent/US10852143B2/en active Active
  • 2019-06-27 WO PCT/CA2019/050897 patent/WO2020000105A1/en not_active Ceased
  • 2019-06-27 JP JP2020572974A patent/JP7390322B2/ja active Active
  • 2019-06-27 MX MX2021000103A patent/MX2021000103A/es unknown
  • 2019-06-27 SG SG11202013082XA patent/SG11202013082XA/en unknown
• 2020
  • 2020-10-29 US US17/084,477 patent/US20210190498A1/en not_active Abandoned
• 2023
  • 2023-07-03 JP JP2023109209A patent/JP7565413B2/ja active Active

Patent Citations (4)

Also Published As

Similar Documents

Legal Events