Systems and methods for motion measurement drift correction

Abstract

This disclosure relates to systems, media, and methods for mitigating measurement drift and improving IMU odometry measurement. In an embodiment, the system may perform operations including receiving first sensor data from at least one motion sensor; receiving 3-dimensional (3-D) motion data based on motion detected by at least one camera; inputting model input data into a machine learning model configured to generate at least one vector, the model input data being based on the received first sensor data and the received 3-D motion data; and apply the at least one vector as an offset.

1 - 20 . (canceled) 
     
     
         21 . A computer-implemented device comprising:
 a memory storing instructions; and   at least one processor configured to execute the instructions to:
 receive first sensor data from at least one sensor; 
 receive motion data; 
 input model input data into a machine learning model configured to generate at least one vector, the model input data being based on the received first sensor data and the received motion data; 
 receive the at least one vector generated by the machine learning model; and 
 apply the at least one vector as an offset to at least one of the received first sensor data or second sensor data. 
   
     
     
         22 . The computer-implemented device of  claim 21 , wherein at least one of the first sensor data or the motion data is received by the computer-implemented device using a Bluetooth® connection. 
     
     
         23 . The computer-implemented device of  claim 21 , wherein the at least one processor is configured to execute an instruction to issue a prompt indicating that data drift has been detected. 
     
     
         24 . The computer-implemented device of  claim 21 , wherein the at least one processor is configured to execute an instruction to display at least one interface allowing for user interaction with at least one of the first sensor data, the motion data, a parameter of the machine learning model, an experiment parameter, a drift offset parameter, or a device calibration setting. 
     
     
         25 . The computer-implemented device of  claim 21 , wherein the machine learning model comprises layers, the layers including at least one of:
 a convolution layer;   a linear layer;   a soft maximum filter;   a drop-out layer;   a batch normalization layer;   a concatenation layer; or   a one-dimensional layer.   
     
     
         26 . The computer-implemented device of  claim 21 , wherein the at least one processor is configured to execute an instruction to select the machine learning model from among a plurality of machine learning models, the selection being based on at least one input received at the computer-implemented device from a user. 
     
     
         27 . The computer-implemented device of  claim 26 , wherein the machine learning models are associated with different contextual attributes, the contextual attributes comprising at least one of:
 a data capture device type;   a motion type;   a sensor placement;   a source of motion;   an environment condition; or   a user identifier.   
     
     
         28 . The computer-implemented device of  claim 21 , wherein the at least one processor is configured to execute an instruction to generate the model input data by forming data points from a data stream into a matrix. 
     
     
         29 . The computer-implemented device of  claim 21 , wherein:
 the at least one processor is configured to execute the instructions to apply the at least one vector as an offset to the second sensor data;   the first sensor data is associated with a first time period;   the second sensor data is associated with a second time period following the first time period.   
     
     
         30 . The computer-implemented device of  claim 21 , wherein the at least one motion sensor comprises an inertial measurement unit (IMU). 
     
     
         31 . The computer-implemented device of  claim 21 , wherein the motion data is received from at least one camera. 
     
     
         32 . The computer-implemented device of  claim 21 , wherein the at least one processor is configured to execute an instruction to apply a data transformation to the received first sensor data or the motion data to generate at least a portion of the model input data. 
     
     
         33 . The computer-implemented device of  claim 32 , wherein the data transformation comprises at least one of:
 a data smoothing operation;   a moving average operation; or   a coordinate frame transformation.   
     
     
         34 . The computer-implemented device of  claim 32 , wherein the data transformation comprises changing an initial format of the received first sensor data or the motion data into a format interpretable by the machine learning model. 
     
     
         35 . The computer-implemented device of  claim 21 , wherein the model input data comprises at least one of:
 a value representing absolute orientation in space;   a value representing a measurement of a gravitational force;   a value representing angular velocity; or   a value representing linear acceleration.   
     
     
         36 . The computer-implemented device of  claim 21 , wherein at least one of the first sensor data or the motion data comprises time series data. 
     
     
         37 . The computer-implemented device of  claim 21 , wherein:
 applying the at least one vector as an offset produces offset data; and   the at least one processor is configured to execute an instruction to provide one or more recommendations, the one or more recommendations based on a comparison of the offset data to reference data.   
     
     
         38 . The computer-implemented device of  claim 21 , wherein:
 the motion data is received from at least one camera; and   the first sensor data and the 3-D motion data are generated while the at least one motion sensor or the at least one camera is in contact with a moving entity.   
     
     
         39 . A method comprising:
 receiving first sensor data from at least one sensor;   receiving motion data;   inputting model input data into a machine learning model configured to generate at least one vector, the model input data being based on the received first sensor data and the received motion data;   receiving the at least one vector generated by the machine learning model; and   applying the at least one vector as an offset to at least one of the received first sensor data or second sensor data.   
     
     
         40 . A non-transitory computer-readable medium storing instructions that, when executed by at least one processor, cause the at least one processor to:
 receive first sensor data from at least one sensor;   receive motion data;   input model input data into a machine learning model configured to generate at least one vector, the model input data being based on the received first sensor data and the received motion data;   receive the at least one vector generated by the machine learning model; and   apply the at least one vector as an offset to at least one of the received first sensor data or second sensor data.