Trajectory and intent prediction
Methods and systems for trajectory and intent prediction are provided. The methods and systems include operations comprising: receiving an observed trajectory of a user and user behavior information; processing the observed trajectory by a machine learning technique to generate a plurality of predicted trajectories, the machine learning technique being trained to establish a relationship between a plurality of training observed trajectories and training predicted trajectories; adjusting the plurality of predicted trajectories based on the user behavior information to determine user intent to operate a target access control device; determining that the target access control device within a threshold range of a given one of the plurality of predicted trajectories; and in response to determining that the target access control device is within the threshold range of the given one of the plurality of predicted trajectories, performing an operation associated with the target access control device.
1. A method comprising:
receiving, by one or more processors, an observed trajectory of a user and user behavior information for the user;
processing the observed trajectory by a machine learning technique to generate a plurality of predicted trajectories, the machine learning technique being trained to establish a relationship between a plurality of training observed trajectories and training predicted trajectories, a first predicted trajectory of the plurality of predicted trajectories representing a first future path the user will follow from the observed trajectory and a second predicted trajectory of the plurality of predicted trajectories representing a second future path the user will follow from the observed trajectory;
adjusting the plurality of predicted trajectories based on the user behavior information to determine user intent to operate a target access control device;
determining that the target access control device is within a threshold range of a given one of the plurality of predicted trajectories; and
in response to determining that the target access control device is within the threshold range of the given one of the plurality of predicted trajectories, performing an operation associated with the target access control device.
2. The method of claim 1 , wherein the target access control device comprises a lock associated with a door, and wherein the performing the operation comprises unlocking the door.
3. The method of claim 2 , further comprising:
establishing a wireless communication link between a mobile device of a user and the target access control device;
exchanging authorization information over the wireless communication link; and
performing the operation after determining that the user is authorized, based on the authorization information, to access the target access control device.
4. The method of claim 3 , further comprising:
determining that the user is authorized, based on the authorization information, to access the target access control device prior to performing the operation; and
delaying performing the operation after determining that the user is authorized until the target access control device is determined to be within the threshold range of the given one of the plurality of predicted trajectories.
5. The method of claim 3 , further comprising:
determining that the user is authorized, based on the authorization information, to access the target access control device prior to performing the operation; and
preventing performing the operation after determining that the user is authorized in response to determining that the target access control device is outside of the threshold range of the given one of the plurality of predicted trajectories.
6. The method of claim 1 , wherein the machine learning technique comprises a conditioned variational autoencoder.
7. The method of claim 6 , wherein adjusting the plurality of predicted trajectories based on the user behavior information comprises processing the observed trajectory and the user behavior information by the conditioned variational autoencoder to generate the plurality of predicted trajectories, wherein each of the plurality of predicted trajectories is associated with a respective probability indicating a likelihood that the user will travel along the corresponding predicted trajectory.
8. The method of claim 1 , wherein the machine learning technique comprises a variational autoencoder.
9. The method of claim 8 , wherein adjusting the plurality of predicted trajectories based on the user behavior information comprise concatenating the user behavior information with the plurality of predicted trajectories output by the variational autoencoder, wherein each of the plurality of predicted trajectories is associated with a respective probability indicating a likelihood that the user will travel along the corresponding predicted trajectory.
10. The method of claim 8 , further comprising processing the concatenated user behavior information and the plurality of predicted trajectories with a second machine learning technique, the second machine learning technique being trained to establish a relationship between a plurality of training user behavior information and predicted intentions of operating access control devices.
11. The method of claim 1 , further comprising encoding the observed trajectory of the user, wherein the machine learning technique is applied to the encoded observed trajectory of the user.
12. The method of claim 1 , further comprising:
determining whether the received user behavior information satisfies a minimum parameter of user behavior information.
13. The method of claim 12 , further comprising:
in response to determining that the received user behavior information satisfies the minimum parameter of user behavior information, allowing the target access control device to perform the operation.
14. The method of claim 12 further comprising:
in response to determining that the received user behavior information fails to satisfy the minimum parameter of user behavior information, preventing the target access control device from performing the operation.
15. The method of claim 12 , wherein the minimum parameter comprises a threshold quantity of specified types of user behavior information.
16. The method of claim 1 , further comprising generating the user behavior information by encoding a feature vector that includes at least one of:
monitoring physical movement of the user;
monitoring a stride of the user;
identifying times and locations at which the user operates different types of access control devices;
identifying other client devices and other types of access control devices within range of the user when a given access control device is being operated by the user; or
identifying other users who are typically in his/her social network.
17. The method of claim 1 , wherein the machine learning technique comprises a first machine learning technique, further comprising:
generating the user behavior information by a second machine learning technique, the second machine learning technique being trained to establish a relationship between training user behavior information and predicted user behavior information; and
generating the user intent to operate the target access control device by a third machine learning technique, the third machine learning technique being trained to establish a relationship between training user behavior information concatenated with a set of trajectories and predicted user intent to operate access control devices.
18. A system comprising:
one or more processors coupled to a memory comprising non-transitory computer instructions that when executed by the one or more processors perform operations comprising:
receiving an observed trajectory of a user and user behavior information for the user;
processing the observed trajectory by a machine learning technique to generate a plurality of predicted trajectories, the machine learning technique being trained to establish a relationship between a plurality of training observed trajectories and training predicted trajectories, a first predicted trajectory of the plurality of predicted trajectories representing a first future path the user will follow from the observed trajectory and a second predicted trajectory of the plurality of predicted trajectories representing a second future path the user will follow from the observed trajectory;
adjusting the plurality of predicted trajectories based on the user behavior information to determine user intent to operate a target access control device;
determining that the target access control device is within a threshold range of a given one of the plurality of predicted trajectories; and
in response to determining that the target access control device is within the threshold range of the given one of the plurality of predicted trajectories, performing an operation associated with the target access control device.
19. A non-transitory computer readable medium comprising non-transitory computer-readable instructions for performing operations comprising:
receiving an observed trajectory of a user and user behavior information for the user;
processing the observed trajectory by a machine learning technique to generate a plurality of predicted trajectories, the machine learning technique being trained to establish a relationship between a plurality of training observed trajectories and training predicted trajectories, a first predicted trajectory of the plurality of predicted trajectories representing a first future path the user will follow from the observed trajectory and a second predicted trajectory of the plurality of predicted trajectories representing a second future path the user will follow from the observed trajectory;
adjusting the plurality of predicted trajectories based on the user behavior information to determine user intent to operate a target access control device;
determining that the target access control device is within a threshold range of a given one of the plurality of predicted trajectories; and
in response to determining that the target access control device is within the threshold range of the given one of the plurality of predicted trajectories, performing an operation associated with the target access control device.
20. The non-transitory computer readable medium of claim 19 , wherein training the machine learning technique comprises:
receiving pairs of training data comprising observed trajectories and corresponding ground-truth trajectories, wherein the ground-truth trajectories represent subsequent trajectories that follow the observed trajectories;
applying model parameters of the machine learning technique to a first batch of the training data to generate estimated predicted trajectories;
computing a derivative of a loss function based on comparing the estimated predicted trajectories to the ground-truth trajectories;
updating the model parameters of the machine learning technique based on the computed derivative of the loss function; and
iteratively applying the updated model parameters to additional batches of training data until meeting a specified convergence criteria.