Abstract: In some implementations, a method includes obtaining a semantic construction of a physical environment. In some implementations, the semantic construction of the physical environment includes a representation of a physical element and a semantic label for the physical element. In some implementations, the method includes obtaining a graphical representation of the physical element. In some implementations, the method includes synthesizing a perceptual property vector (PPV) for the graphical representation of the physical element based on the semantic label for the physical element. In some implementations, the PPV includes one or more perceptual characteristic values characterizing the graphical representation of the physical element. In some implementations, the method includes compositing an affordance in association with the graphical representation of the physical element.

Abstract: In one implementation, a method of defining a negative space in a three-dimensional scene model is performed at a device including a processor and non-transitory memory. The method includes obtaining a three-dimensional scene model of a physical environment including a plurality of points, wherein each of the plurality of points is associated with a set of coordinates in a three-dimensional space. The method includes defining a subspace in the three-dimensional space with less than a threshold number of the plurality of points. The method includes determining a semantic label for the subspace. The method includes generating a characterization vector of the subspace, wherein the characterization vector includes the spatial extent of the subspace and the semantic label.

Abstract: In some implementations, a method includes obtaining environmental data corresponding to a physical environment. In some implementations, the method includes determining, based on the environmental data, a bounding surface of the physical environment. In some implementations, the method includes detecting a physical element located within the physical environment based on the environmental data. In some implementations, the method includes determining a semantic label for the physical element based on at least a portion of the environmental data corresponding to the physical element. In some implementations, the method includes generating a semantic construction of the physical environment based on the environmental data. In some implementations, the semantic construction of the physical environment includes a representation of the bounding surface, a representation of the physical element and the semantic label for the physical element.

Abstract: Various implementations disclosed herein include devices, systems, and methods for granular motion control for a virtual agent. In various implementations, a device includes a non-transitory memory and one or more processors coupled with the non-transitory memory. In some implementations, a method includes obtaining an action for a virtual agent. In some implementations, the action is associated with a plurality of time frames. In some implementations, the method includes, for a first time frame of the plurality of time frames, determining respective confidence scores for a plurality of granular motions that advance the virtual agent towards completion of the action. In some implementations, the method includes selecting a subset of the plurality of granular motions based on the respective confidence scores.

Abstract: In one implementation, a method for modeling a behavior with synthetic training data. The method includes: obtaining source content that includes an entity performing one or more actions within an environment; generating a first environment characterization vector characterizing the environment; generating a first set of behavioral trajectories associated with the one or more actions of the entity based on the source content and the first characterization vector for the environment; generating a second environment characterization vector for the environment by perturbing the first environment characterization vector; generating a second set of behavioral trajectories associated with one or more potential actions of the entity based on the source content and the second characterization vector for the environment; and training a behavior model for a virtual agent based on the first and second sets of behavioral trajectories in order to imitate the entity.

Abstract: In one implementation, a method of defining a negative space in a three-dimensional scene model is performed at a device including a processor and non-transitory memory. The method includes obtaining a three-dimensional scene model of a physical environment including a plurality of points, wherein each of the plurality of points is associated with a set of coordinates in a three-dimensional space. The method includes defining a subspace in the three-dimensional space with less than a threshold number of the plurality of points. The method includes determining a semantic label for the subspace. The method includes generating a characterization vector of the subspace, wherein the characterization vector includes the spatial extent of the subspace and the semantic label.

Abstract: Various implementations disclosed herein include devices, systems, and methods for granular motion control for a virtual agent. In various implementations, a device includes a non-transitory memory and one or more processors coupled with the non-transitory memory. In some implementations, a method includes obtaining an action for a virtual agent. In some implementations, the action is associated with a plurality of time frames. In some implementations, the method includes, for a first time frame of the plurality of time frames, determining respective confidence scores for a plurality of granular motions that advance the virtual agent towards completion of the action. In some implementations, the method includes selecting a subset of the plurality of granular motions based on the respective confidence scores.

Abstract: In one implementation, a method for generating a blended animation. The method includes: obtaining a motion input vector for a current time period; generating a motion output vector and pose information for the current time period based on the motion input vector; selecting an animated motion from a bank of animated motions for the current time period that matches the pose information within a threshold tolerance value; obtaining a blending coefficients vector for the current time period; generating a blended animation for the current time period by blending the motion output vector with the animated motion based on the blending coefficients vector; and generating a reward signal for the blended animation for the current time period.

Abstract: In one implementation, a method of defining a negative space in a three-dimensional scene model is performed at a device including a processor and non-transitory memory. The method includes obtaining a three-dimensional scene model of a physical environment including a plurality of points, wherein each of the plurality of points is associated with a set of coordinates in a three-dimensional space. The method includes defining a subspace in the three-dimensional space with less than a threshold number of the plurality of points. The method includes determining a semantic label for the subspace. The method includes generating a characterization vector of the subspace, wherein the characterization vector includes the spatial extent of the subspace and the semantic label.

Abstract: In one implementation, a method for inverse reinforcement learning for tailoring virtual agent behaviors to a specific user. The method includes: obtaining an initial behavior model for a virtual agent and an initial state for a virtual environment associated with the virtual agent, wherein the initial behavior model includes one or more tunable parameters; generating, based on the initial behavior model and the initial state for the virtual environment, a first set of behavioral trajectories for the virtual agent; obtaining a second set of behavioral trajectories from a source different from the initial behavior model; and generating an updated behavior model by adjusting at least one of the one or more tunable parameters of the initial behavior model as a function of the first and second sets of behavioral trajectories, wherein at least one of the first and second sets of behavioral trajectories are assigned different weights.

Abstract: In one implementation, a method for generating a blended animation. The method includes: obtaining a motion input vector for a current time period; generating a motion output vector and pose information for the current time period based on the motion input vector; selecting an animated motion from a bank of animated motions for the current time period that matches the pose information within a threshold tolerance value; obtaining a blending coefficients vector for the current time period; generating a blended animation for the current time period by blending the motion output vector with the animated motion based on the blending coefficients vector; and generating a reward signal for the blended animation for the current time period.

Abstract: In one implementation, a method for generating a blended animation. The method includes: obtaining a motion input vector for a current time period; generating a motion output vector and pose information for the current time period based on the motion input vector; selecting an animated motion from a bank of animated motions for the current time period that matches the pose information within a threshold tolerance value; obtaining a blending coefficients vector for the current time period; generating a blended animation for the current time period by blending the motion output vector with the animated motion based on the blending coefficients vector; and generating a reward signal for the blended animation for the current time period.

Abstract: A method includes generating, in coordination with an emergent content engine, a first objective for a first objective-effectuator and a second objective for a second objective-effectuator instantiated in a computer-generated reality (CGR) environment. The first and second objectives are associated with a mutual plan. The method includes generating, based on characteristic values associated with the first and second objective-effectuators a first directive for the first objective-effectuator and a second directive for the second objective-effectuator. The first directive limits actions generated by the first objective-effectuator over a first set of time frames associated with the first objective and the second directive limits actions generated by the second objective-effectuator over a second set of time frames associated with the second objective.

Abstract: A method includes obtaining video data including a representation of a first plurality of motions of a real-world entity. The method includes determining, from the video data, a plurality of estimated torque values using a motion controller. The plurality of estimated torque values is associated with a plurality of real-world joints corresponding to the first plurality of motions. The method includes generating a second plurality of motions of a virtual agent by providing the plurality of estimated torque values to a corresponding plurality of virtual joints of the virtual agent. Movement of the virtual agent is controllable by the plurality of virtual joints, corresponding to the plurality of real-world joints of the real-world entity. The method includes, responsive to a determination that a comparison between the first plurality of motions and the second plurality of motions does not satisfy a performance metric, changing an operational value of the motion controller.

Abstract: In some implementations, a method includes obtaining an objective for a computer-generated reality (CGR) representation of an objective-effectuator. In some implementations, the objective is associated with a plurality of time frames. In some implementations, the method includes determining a plurality of candidate plans that satisfy the objective. In some implementations, the method includes selecting a first candidate plan of the plurality of candidate plans based on a selection criterion. In some implementations, the method includes effectuating the first candidate plan in order to satisfy the objective. In some implementations, the first candidate plan triggers the CGR representation of the objective-effectuator to perform a series of actions over the plurality of time frames associated with the objective.

Abstract: In some implementations, a method includes obtaining an objective for a computer-generated reality (CGR) representation of an objective-effectuator. In some implementations, the objective is associated with a plurality of time frames. In some implementations, the method includes determining a plurality of candidate plans that satisfy the objective. In some implementations, the method includes selecting a first candidate plan of the plurality of candidate plans based on a selection criterion. In some implementations, the method includes effectuating the first candidate plan in order to satisfy the objective. In some implementations, the first candidate plan triggers the CGR representation of the objective-effectuator to perform a series of actions over the plurality of time frames associated with the objective.

Abstract: In some implementations, a method includes obtaining environmental data corresponding to a physical environment. In some implementations, the method includes determining, based on the environmental data, a bounding surface of the physical environment. In some implementations, the method includes detecting a physical element located within the physical environment based on the environmental data. In some implementations, the method includes determining a semantic label for the physical element based on at least a portion of the environmental data corresponding to the physical element. In some implementations, the method includes generating a semantic construction of the physical environment based on the environmental data. In some implementations, the semantic construction of the physical environment includes a representation of the bounding surface, a representation of the physical element and the semantic label for the physical element.

Abstract: In some implementations, a method includes obtaining a semantic construction of a physical environment. In some implementations, the semantic construction of the physical environment includes a representation of a physical element and a semantic label for the physical element. In some implementations, the method includes obtaining a graphical representation of the physical element. In some implementations, the method includes synthesizing a perceptual property vector (PPV) for the graphical representation of the physical element based on the semantic label for the physical element. In some implementations, the PPV includes one or more perceptual characteristic values characterizing the graphical representation of the physical element. In some implementations, the method includes compositing an affordance in association with the graphical representation of the physical element.

Abstract: In some implementations, a method includes obtaining, by a virtual intelligent agent (VIA), a perceptual property vector (PPV) for a graphical representation of a physical element. In some implementations, the PPV includes one or more perceptual characteristic values characterizing the graphical representation of the physical element. In some implementations, the method includes instantiating a graphical representation of the VIA in a graphical environment that includes the graphical representation of the physical element and an affordance that is associated with the graphical representation of the physical element. In some implementations, the method includes generating, by the VIA, an action for the graphical representation of the VIA based on the PPV. In some implementations, the method includes displaying a manipulation of the affordance by the graphical representation of the VIA in order to effectuate the action generated by the VIA.

Abstract: A method includes determining a first portion of state information that is accessible to a first agent instantiated in an environment. The method includes determining a second portion of the state information that is accessible to a second agent instantiated in the environment. The method includes generating a first set of actions for a representation of the first agent based on the first portion of the state information to satisfy a first objective of the first agent. The method includes generating a second set of actions for a representation of the second agent based on the second portion of the state information to satisfy a second objective of the second agent. The method includes modifying the representations of the first and second agents based on the first and second set of actions.