Utilizing context information of environment component regions for event/activity prediction

Abstract

Sensor data is received from a physical environment sensor. A component region associated with at least a portion of the sensor data is identified. A physical environment has been defined to include a plurality of component regions. Context information associated with the identified component region is obtained. The context information is utilized in association with a prediction model.

What is claimed is: 
     
         1 - 21 . (canceled) 
     
     
         22 . A computer-executed method for home automation, comprising:
 determining, by a computer, one or more component regions within a physical environment;   determining, for a respective component region, context information associated with the component region;   receiving sensor data from one or more sensors in the physical environment; and   applying a prediction model to determine a state of a detected subject in the physical environment based on the determined component regions, context information associated with the component regions, and the received sensor data.   
     
     
         23 . The method of  claim 22 , wherein applying the prediction model further comprises:
 identifying one or more component regions associated with the received sensor data;   selecting one or more signal aggregators based on context information associated with the identified component regions; and   aggregating, by at least one selected signal aggregator, sensor data from different sensors and corresponding context information to obtain a contextual input for the prediction model.   
     
     
         24 . The method of  claim 23 , wherein a respective signal aggregator comprises one or more of:
 a neural network;   an aggregation function;   a rule function; and   a machine-learning model trained across multiple physical environments.   
     
     
         25 . The method of  claim 22 , further comprising:
 in response to determining that multiple sensors monitor a same component region, correlating sensor data provided by the multiple sensors together.   
     
     
         26 . The method of  claim 22 , wherein the one or more sensors in the physical environment include at least one of the following: a switch, a camera, a motion detector, an infrared detector, a light detector, an accelerometer, a thermal detector, a human pose detector, a thermometer, an air quality sensor, a smoke detector, a microphone, a humidity detector, a door sensor, a window sensor, a water detector, a glass breakage detector, and a power utilization sensor. 
     
     
         27 . The method of  claim 22 , wherein the component regions include a hierarchy of component regions. 
     
     
         28 . The method of  claim 22 , wherein the component regions have been defined, at least in part, using a graph model of the physical environment. 
     
     
         29 . The method of  claim 28 , wherein determining the context information associated with the component region comprises determining the context information from the graph model of the physical environment. 
     
     
         30 . The method of  claim 22 , wherein the context information includes one or more of: spatial relationship information associated with the component region, functional information associated with the component region, information about an item associated with the component region, and information about functional capabilities of the item associated with the component region. 
     
     
         31 . The method of  claim 22 , wherein determining the state of the detected subject comprises determining an activity performed by a human user. 
     
     
         32 . A computer system, comprising:
 a processer;   a storage device coupled to the processor and storing instructions, which when executed by the processor cause the processor to perform a method for home automation, the method comprising:
 determining, by a computer, one or more component regions within a physical environment; 
 determining, for a respective component region, context information associated with the component region; 
 receiving sensor data from one or more sensors in the physical environment; and 
 applying a prediction model to determine a state of a detected subject based on the determined component regions, context information associated with the component regions, and the received sensor data. 
   
     
     
         33 . The computer system of  claim 32 , wherein applying the prediction model further comprises:
 identifying one or more component regions associated with the received sensor data;   selecting one or more signal aggregators based on context information associated with the identified component regions; and   aggregating, by at least one selected signal aggregator, sensor data from different sensors and corresponding context information to obtain a contextual input for the prediction model.   
     
     
         34 . The computer system of  claim 33 , wherein a respective signal aggregator comprises one or more of:
 a neural network;   an aggregation function;   a rule function; and   a machine-learning model trained across multiple physical environments.   
     
     
         35 . The computer system of  claim 32 , further comprising:
 in response to determining that multiple sensors monitor a same component region, correlating sensor data provided by the multiple sensors together.   
     
     
         36 . The computer system of  claim 32 , wherein the one or more sensors in the physical environment include at least one of the following: a switch, a camera, a motion detector, an infrared detector, a light detector, an accelerometer, a thermal detector, a human pose detector, a thermometer, an air quality sensor, a smoke detector, a microphone, a humidity detector, a door sensor, a window sensor, a water detector, a glass breakage detector, and a power utilization sensor. 
     
     
         37 . The computer system of  claim 32 , wherein the component regions include a hierarchy of component regions. 
     
     
         38 . The computer system of  claim 32 , wherein the component regions have been defined, at least in part, using a graph model of the physical environment. 
     
     
         39 . The computer system of  claim 38 , wherein determining the context information associated with the component region comprises determining the context information from the graph model of the physical environment. 
     
     
         40 . The computer system of  claim 32 , wherein the context information includes one or more of: spatial relationship information associated with the component region, functional information associated with the component region, information about an item associated with the component region, and information about functional capabilities of the item associated with the component region. 
     
     
         41 . The computer system of  claim 32 , wherein determining the state of the detected subject comprises determining an activity performed by a human user.