Assessing present intentions of an actor perceived by an autonomous vehicle

Abstract

Systems and methods for controlling an autonomous vehicle (AV). The methods comprise: generating candidate intentions of an actor based on a detected action of the actor and a classification associated with the actor; determining an overall probability for each candidate intention based on at least a persistence of the candidate intention over a non-interrupted sequence of cycles (where each cycle represents a time period over which the actor was sensed by a sensor); selecting candidate intention(s) based on the overall probabilities; forecasting a subsequent future intention that the actor may have after reaching a goal defined by the candidate intention(s) which was(were) selected; obtaining an actor trajectory that is consistent with the candidate intention(s) which was(were) selected and the subsequent future intention; and using the actor trajectory to influence a selected trajectory for AV.

1 . A method for controlling an autonomous vehicle, comprising:
 generating, by a processor, a plurality of candidate intentions of an actor based on a detected action of the actor and a classification associated with the actor;   determining, by the processor, an overall probability for each candidate intention of the plurality of candidate intentions based on at least a persistence of the candidate intention over a non-interrupted sequence of cycles, where each one of said cycles represents a time period over which the actor was sensed by a sensor;   selecting, by the processor, at least one candidate intention of the plurality of candidate intentions based on the overall probabilities;   forecasting, by the processor, a subsequent future intention that the actor may have after reaching a goal defined by the at least one candidate intention which was selected;   obtaining, by the processor, an actor trajectory that is consistent with the at least one candidate intention which was selected and the subsequent future intention; and   using the actor trajectory to influence a selected trajectory for the autonomous vehicle.   
     
     
         2 . The method according to  claim 1 , wherein the generating the plurality of candidate intentions of the actor comprises:
 accessing a data set of possible goals that are associated with various classes of actors;   selecting possible goals that the data set associates with the detected class of the actor;   determining which of the possible goals in the data set are consistent with the detected action; and   using the determined possible goals as the candidate intentions.   
     
     
         3 . The method according to  claim 2 , wherein the determining which of the possible goals in the data set are consistent with the detected action comprises:
 determining whether the detected action satisfies one or more rules of each of the possible goals; or   processing the detected actions and possible goals in a machine learning model that has been trained on a data set of labeled actions and goals.   
     
     
         4 . The method according  claim 1 , further comprising refining the overall probability for each candidate intention based on likelihoods of candidate intentions generated for the actor in next cycles. 
     
     
         5 . The method according to  claim 1 , further comprising:
 assigning a probability to the subsequent future-intention that the actor may have; and   comparing the probability to a threshold;   wherein the obtaining the actor trajectory is performed responsive to a results of said comparing.   
     
     
         6 . The method according to  claim 5 , wherein said forecasting the future subsequent intention using one or more of the plurality of candidate intentions which have likelihoods above a threshold. 
     
     
         7 . The method according to  claim 1 , wherein determining the overall probability for each candidate intention comprises:
 assigning a first likelihood to a candidate intention that persisted over the non-interrupted sequence of cycles, that is consistent with the kinematic state of the actor and/or that is consistent with an environmental condition; and   assigning a different second likelihood to a candidate intention that did not persist over the non-interrupted sequence of cycles, that is not consistent with the kinematic state of the actor, and/or that is not consistent with the environmental condition.   
     
     
         8 . A system, comprising:
 a processor, and   a non-transitory computer-readable storage medium comprising programming instructions that are configured to cause the processor to implement a method for controlling an autonomous vehicle, wherein the programming instructions comprise instructions to:
 generate a plurality of candidate intentions of an actor based on a detected action of the actor and a classification associated with the actor; 
 determine an overall probability for each candidate intention of the plurality of candidate intentions based on at least a persistence of the candidate intention over a non-interrupted sequence of cycles, where each one of said cycles represents a time period over which the actor was sensed by a sensor; 
 select at least one candidate intention of the plurality of candidate intentions based on the overall probabilities; 
 forecast a subsequent future intention that the actor may have after reaching a goal defined by the at least one candidate intention which was selected; 
 obtain an actor trajectory that is consistent with the at least one candidate intention which was selected and the subsequent future intention; and 
 use the actor trajectory to influence a selected trajectory for the autonomous vehicle. 
   
     
     
         9 . The system according to  claim 8 , wherein the plurality of candidate intentions of the actor are generated by:
 accessing a data set of possible goals that are associated with various classes of actors;   selecting possible goals that the data set associates with the detected class of the actor;   determining which of the possible goals in the data set are consistent with the detected action; and   using the determined possible goals as the candidate intentions.   
     
     
         10 . The system according to  claim 9 , wherein the possible goals are selected by:
 determining whether the detected action satisfies one or more rules of each of the possible goals; or   processing the detected actions and possible goals in a machine learning model that has been trained on a data set of labeled actions and goals.   
     
     
         11 . The system according  claim 8 , wherein the programming instructions further comprise instructions to refine the overall probability for each candidate intention based on likelihoods of candidate intentions generated for the actor in next cycles. 
     
     
         12 . The system according to  claim 8 , wherein the programming instructions further comprise instructions to:
 assign a probability to the subsequent future-intention that the actor may have; and   compare the probability to a threshold;   wherein the obtaining the actor trajectory is performed responsive to a results of said comparing.   
     
     
         13 . The system according to  claim 12 , wherein the future subsequent intention is forecasted by using one or more of the plurality of candidate intentions which have likelihoods above a threshold. 
     
     
         14 . The system according to  claim 8 , wherein the overall probability for each candidate intention is determined by:
 assigning a first likelihood to a candidate intention that persisted over the non-interrupted sequence of cycles, that is consistent with the kinematic state of the actor and/or that is consistent with an environmental condition; and   assigning a different second likelihood to a candidate intention that did not persist over the non-interrupted sequence of cycles, that is not consistent with the kinematic state of the actor, and/or that is not consistent with the environmental condition.   
     
     
         15 . A non-transitory computer-readable medium that stores instructions that are configured to, when executed by at least one computing device, cause the at least one computing device to perform operations comprising:
 generating a plurality of candidate intentions of an actor based on a detected action of the actor and a classification associated with the actor;   determining an overall probability for each candidate intention of the plurality of candidate intentions based on at least a persistence of the candidate intention over a non-interrupted sequence of cycles, where each one of said cycles represents a time period over which the actor was sensed by a sensor;   selecting at least one candidate intention of the plurality of candidate intentions based on the overall probabilities;   forecasting a subsequent future intention that the actor may have after reaching a goal defined by the at least one candidate intention which was selected;   obtaining an actor trajectory that is consistent with the at least one candidate intention which was selected and the subsequent future intention; and   using the actor trajectory to influence a selected trajectory for an autonomous vehicle.   
     
     
         16 . The non-transitory computer-readable medium according to  claim 15 , wherein the generating the plurality of candidate intentions of the actor comprises:
 accessing a data set of possible goals that are associated with various classes of actors;   selecting possible goals that the data set associates with the detected class of the actor;   determining which of the possible goals in the data set are consistent with the detected action; and   using the determined possible goals as the candidate intentions.   
     
     
         17 . The non-transitory computer-readable medium according to  claim 16 , wherein the determining which of the possible goals in the data set are consistent with the detected action comprises:
 determining whether the detected action satisfies one or more rules of each of the possible goals; or   processing the detected actions and possible goals in a machine learning model that has been trained on a data set of labeled actions and goals.   
     
     
         18 . The non-transitory computer-readable medium according  claim 15 , wherein the at least one computing device is further caused to refine the overall probability for each candidate intention based on likelihoods of candidate intentions generated for the actor in next cycles. 
     
     
         19 . The non-transitory computer-readable medium according to  claim 15 , wherein the at least one computing device is further caused to:
 assign a probability to the subsequent future-intention that the actor may have; and   compare the probability to a threshold;   wherein the obtaining the actor trajectory is performed responsive to a results of said comparing.   
     
     
         20 . The non-transitory computer-readable medium according to  claim 15 , wherein said forecasting the future subsequent intention using one or more of the plurality of candidate intentions which have likelihoods above a threshold.