Bending estimation as a biometric signal
A method for generating reference biometric data based on a bending of a flexible device is described. In one aspect, a method includes forming training data includes bending estimates of a flexible device worn by a first user, training a model based on the training data, and generating reference biometric data for the first user based on the model.
1 . A method comprising: forming training data comprising bending estimates of a flexible device worn by a first user, the bending estimates indicating how deformed the flexible device is and being derived from physiological characteristics of the first user; training a model based on the training data using a machine learning algorithm configured to transform the bending estimates into a biometric profile unique to the first user; and generating reference biometric data for the first user based on the model; wherein the flexible device is head-worn and comprises: a left temple, a right temple, and a frame, wherein the bending estimates comprise: a bending of the left temple with respect to the frame or the right temple; and a bending of the right temple with respect to the frame or the left temple; wherein the bending estimates are based on comparing a left image from a left camera mounted on the left temple with a right image from a right camera mounted on the right temple, VIO data of the flexible device, and a depth map based on the left image and the right image.
2 . The method of claim 1 , further comprising: estimating a bending of the flexible device over a plurality of sessions, each session comprising a wearing of the flexible device by the first user; and forming the bending estimates based on the bending over the plurality of sessions.
3 . The method of claim 1 , wherein the reference biometric data indicate a range of acceptable bending estimates for the first user, wherein the acceptable range is defined as a preset number of degrees from an established baseline bending measurement calibrated to the first user's physiological geometry, and wherein the acceptable range is determined using a calibration process that accounts for the first user's head size, shape, and other physiological characteristics.
4 . The method of claim 1 , further comprising: detecting a wearing of the flexible device by a second user, and an operation of an augmented reality (AR) application of an account associated with the first user at the flexible device; estimating a bending of the flexible device worn by the second user; authenticating the second user based on the bending estimate of the second user and the biometric data of the first user; and in response to the second user being authenticated, granting the second user, access to the AR application of the account associated with the first user.
5 . The method of claim 1 , further comprising: detecting a wearing of the flexible device by a second user, and an operation of an augmented reality (AR) application of an account associated with the first user at the flexible device; estimating a bending of the flexible device worn by the second user; determining that the bending estimate of the flexible device worn by the second user does not match the reference biometric data of the first user; and denying the second user, access to the AR application of the account associated with the first user.
6 . The method of claim 1 , further comprising: detecting a wearing of the flexible device by a second user, and an operation of an augmented reality (AR) application of an account associated with the first user at the flexible device; estimating a bending of the flexible device worn by the second user; determining that the bending estimate of the flexible device worn by the second user does not match the biometric data of the first user; retrieving additional biometric data of the second user based on sensors of the flexible device; and authenticating the second user based on the additional biometric data of the second user and additional reference biometric data of the first user.
7 . The method of claim 6 , wherein the additional biometric data comprises at least one of voice-based biometric data, an iris-based biometric data, or a facial-based biometric data.
8 . The method of claim 1 , wherein the bending estimates comprise combination of a pitch-roll bending estimate and a yaw bending estimate.
9 . A computing apparatus comprising: a processor; and a memory storing instructions that, when executed by the processor, configure the apparatus to perform operations comprising: forming training data comprising bending estimates of a flexible device worn by a first user, the bending estimates indicating how deformed the flexible device is and being derived from physiological characteristics of the first user; training a model based on the training data using a machine learning algorithm configured to transform the bending estimates into a biometric profile unique to the first user; and generating reference biometric data for the first user based on the model wherein the flexible device is head-worn and comprises: a left temple, a right temple, and a frame, wherein the bending estimates comprise: a bending of the left temple with respect to the frame or the right temple; and a bending of the right temple with respect to the frame or the left temple; wherein the bending estimates are based on comparing a left image from a left camera mounted on the left temple with a right image from a right camera mounted on the right temple, VIO data of the flexible device, and a depth map based on the left image and the right image.
10 . The computing apparatus of claim 9 , wherein the instructions further configure the apparatus to: estimate a bending of the flexible device over a plurality of sessions, each session comprising a wearing of the flexible device by the first user; and form the bending estimates based on the bending over the plurality of sessions.
11 . The computing apparatus of claim 9 , wherein the reference biometric data indicate a range of acceptable bend estimates for the first user, wherein the acceptable range is defined as a preset number of degrees from an established baseline bending measurement calibrated to the first user's physiological geometry, and wherein the acceptable range is determined using a calibration process that accounts for the first user's head size, shape, and other physiological characteristics.
12 . The computing apparatus of claim 9 , wherein the instructions further configure the apparatus to: detect a wearing of the flexible device by a second user, and an operation of an augmented reality (AR) application of an account associated with the first user at the flexible device; estimate a bending of the flexible device worn by the second user; authenticate the second user based on the bending estimate of the second user and the biometric data of the first user; and in response to the second user being authenticated, grant the second user, access to the AR application of the account associated with the first user.
13 . The computing apparatus of claim 9 , wherein the instructions further configure the apparatus to: detect a wearing of the flexible device by a second user, and an operation of an augmented reality (AR) application of an account associated with the first user at the flexible device; estimate a bending of the flexible device worn by the second user; determine that the bending estimate of the flexible device worn by the second user does not match the reference biometric data of the first user; and deny the second user, access to the AR application of the account associated with the first user.
14 . The computing apparatus of claim 9 , wherein the instructions further configure the apparatus to:
detect a wearing of the flexible device by a second user, and an operation of an augmented reality (AR) application of an account associated with the first user at the flexible device;
estimate a bending of the flexible device worn by the second user;
determine that the bending estimate of the flexible device worn by the second user does not match the biometric data of the first user;
retrieve additional biometric data of the second user based on sensors of the flexible device; and
authenticate the second user based on the additional biometric data of the second user and additional reference biometric data of the first user.
15 . The computing apparatus of claim 14 , wherein the additional biometric data comprises at least one of voice-based biometric data, an iris-based biometric data, or a facial-based biometric data.
16 . The computing apparatus of claim 9 , wherein the bending estimates comprise combination of a pitch-roll bend estimate and a yaw bending estimate.
17 . A non-transitory computer-readable storage medium, the computer-readable storage medium including instructions that when executed by a computer, cause the computer to perform operations comprising: forming training data comprising bending estimates of a flexible device worn by a first user, the bending estimates indicating how deformed the flexible device is and being derived from physiological characteristics of the first user; training a model based on the training data using a machine learning algorithm configured to transform the bending estimates into a biometric profile unique to the first user; and generating reference biometric data for the first user based on the model wherein the flexible device is head-worn and comprises: a left temple, a right temple, and a frame, wherein the bending estimates comprise: a bending of the left temple with respect to the frame or the right temple; and a bending of the right temple with respect to the frame or the left temple; wherein the bending estimates are based on comparing a left image from a left camera mounted on the left temple with a right image from a right camera mounted on the right temple, VIO data of the flexible device, and a depth map based on the left image and the right image.