Driver behavior risk assessment and pedestrian awareness

Abstract

Driver behavior risk assessment and pedestrian awareness may include an receiving an input stream of images of an environment including one or more objects within the environment, estimating an intention of an ego vehicle based on the input stream of images and a temporal recurrent network (TRN), generating a scene representation based on the input stream of images and a graph neural network (GNN), generating a prediction of a situation based on the scene representation and the intention of the ego vehicle, and generating an influenced or non-influenced action determination based on the prediction of the situation and the scene representation.

1 . A system for driver behavior risk assessment and pedestrian awareness, comprising:
 an intention estimator estimating an intention of an ego vehicle based on an input stream of images of an environment including one or more objects within the environment and a first neural network;   a scene representation generator generating a scene representation based on the input stream of images and a second neural network;   a situation predictor generating a prediction of a situation based on the scene representation and the intention of the ego vehicle; and   a driver response determiner generating an influenced or non-influenced action determination based on the prediction of the situation and the scene representation.   
     
     
         2 . The system for driver behavior risk assessment and pedestrian awareness of  claim 1 , wherein the intention of the ego vehicle is estimated as a left-turn intention, a right-turn intention, or a straight-travel intention. 
     
     
         3 . The system for driver behavior risk assessment and pedestrian awareness of  claim 1 , wherein the environment includes a straight topology, a three-way intersection topology, or a four-way intersection topology. 
     
     
         4 . The system for driver behavior risk assessment and pedestrian awareness of  claim 1 , wherein the situation includes a stop sign, a traffic light, a crossing pedestrian, a crossing vehicle, a vehicle blocking ego lane, a congestion, a jaywalking, a vehicle backing into parking space, a vehicle on shoulder open door, or a cut-in. 
     
     
         5 . The system for driver behavior risk assessment and pedestrian awareness of  claim 1 , comprising a risk object identifier (ROI) extracting image-level and object-level features from the input stream of images of the environment. 
     
     
         6 . The system for driver behavior risk assessment and pedestrian awareness of  claim 5 , wherein the ROI determines one or more object bounding boxes for one or more of the objects within the environment. 
     
     
         7 . The system for driver behavior risk assessment and pedestrian awareness of  claim 6 , wherein one or more of the object bounding boxes are around a face or a head of a pedestrian and the ROI determines whether the pedestrian is looking or is not looking at the ego vehicle. 
     
     
         8 . The system for driver behavior risk assessment and pedestrian awareness of  claim 1 , wherein the situation predictor generates the prediction of the situation based on an element-wise dot product of the scene representation and the intention of the ego vehicle. 
     
     
         9 . The system for driver behavior risk assessment and pedestrian awareness of  claim 1 , wherein the driver response determiner generates the influenced or non-influenced action determination based on passing the prediction of the situation through a multilayer perceptron (MLP) and the scene representation. 
     
     
         10 . The system for driver behavior risk assessment and pedestrian awareness of  claim 1 , wherein the situation predictor classifies the prediction of the situation into a binary class. 
     
     
         11 . A method for driver behavior risk assessment and pedestrian awareness, comprising:
 estimating an intention of an ego vehicle based on an input stream of images of an environment including one or more objects within the environment and a first neural network;   generating a scene representation based on the input stream of images and a second neural network;   generating a prediction of a situation based on the scene representation and the intention of the ego vehicle; and   generating an influenced or non-influenced action determination based on the prediction of the situation and the scene representation.   
     
     
         12 . The method for driver behavior risk assessment and pedestrian awareness of  claim 11 , wherein the intention of the ego vehicle is estimated as a left-turn intention, a right-turn intention, or a straight-travel intention. 
     
     
         13 . The method for driver behavior risk assessment and pedestrian awareness of  claim 11 , wherein the environment includes a straight topology, a three-way intersection topology, or a four-way intersection topology. 
     
     
         14 . The method for driver behavior risk assessment and pedestrian awareness of  claim 11 , wherein the situation includes a stop sign, a traffic light, a crossing pedestrian, a crossing vehicle, a vehicle blocking ego lane, a congestion, a jaywalking, a vehicle backing into parking space, a vehicle on shoulder open door, or a cut-in. 
     
     
         15 . The method for driver behavior risk assessment and pedestrian awareness of  claim 11 , comprising extracting image-level and object-level features from the input stream of images of the environment. 
     
     
         16 . The method for driver behavior risk assessment and pedestrian awareness of  claim 15 , comprising determining one or more object bounding boxes for one or more of the objects within the environment. 
     
     
         17 . The method for driver behavior risk assessment and pedestrian awareness of  claim 16 , comprising determining whether a pedestrian is looking or is not looking at the ego vehicle, wherein one or more of the object bounding boxes are around a face or a head of the pedestrian. 
     
     
         18 . The method for driver behavior risk assessment and pedestrian awareness of  claim 11 , comprising generating the prediction of the situation based on an element-wise dot product of the scene representation and the intention of the ego vehicle. 
     
     
         19 . The method for driver behavior risk assessment and pedestrian awareness of  claim 11 , comprising generating the influenced or non-influenced action determination based on passing the prediction of the situation through a multilayer perceptron (MLP) and the scene representation. 
     
     
         20 . A driver behavior risk assessment and pedestrian awareness vehicle, comprising:
 an intention estimator estimating an intention of the vehicle based on an input stream of images of an environment including one or more objects within the environment and a first neural network;   a scene representation generator generating a scene representation based on the input stream of images and a second neural network;   a situation predictor generating a prediction of a situation based on the scene representation and the intention of the vehicle; and   a driver response determiner generating an influenced or non-influenced action determination based on the prediction of the situation and the scene representation.