Generating simplified object models to reduce computational resource requirements for autonomous vehicles

Abstract

Aspects of the disclosure relate to controlling a vehicle using a simplified model of an object. In one example, sensor data including a plurality of data points corresponding to surfaces of the object in the vehicle's environment may be received from one or more sensors of the vehicle. A first model may be determined using a subset of the plurality of data points. A set of secondary data points may be identified from the plurality of data points using a point on the vehicle. The set of secondary data points may be filtered from the subset of the plurality data points to determine a second model, wherein the second model is a simplified version of the first model. The vehicle may be controlled in an autonomous driving mode based on the second model.

1 . A method of controlling a vehicle using a simplified model of an object, the method comprising:
 receiving, by one or more computing devices from one or more sensors of the vehicle, sensor data including a plurality of three-dimensional data points corresponding to surfaces of the object in the vehicle's environment;   determining, by the one or more computing devices, a first model representative of the object using only two-dimensional data points of the plurality of three-dimensional data points;   generating, by the one or more computing devices, a second model representative of the object by removing data points of the first model corresponding to surfaces that were occluded from the one or more sensors of the vehicle when the sensor data was captured; and   controlling, by the one or more computing devices, the vehicle in an autonomous driving mode based on the second model.   
     
     
         2 . The method of  claim 1 , wherein the plurality of three-dimensional data points are received from a LIDAR sensor of the vehicle. 
     
     
         3 . The method of  claim 1 , wherein the first model is determined by ignoring a height dimension of the plurality of three-dimensional data points. 
     
     
         4 . The method of  claim 1 , wherein the first model is determined by projecting the plurality of three-dimensional data points onto a two-dimensional plane. 
     
     
         5 . The method of  claim 1 , wherein each of the first model and the second model are two-dimensional representations of the object. 
     
     
         6 . The method of  claim 1 , wherein determining the first model includes determining a convex hull from the two-dimensional data points. 
     
     
         7 . The method of  claim 6 , wherein removing data points of the first model includes identifying data points of the convex hull corresponding to a side of the convex hull that are oriented away from the vehicle relative to other data points of the convex hull. 
     
     
         8 . The method of  claim 7 , wherein removing data points of the first model includes determining a modified convex hull based on the two-dimensional data points and a point on the vehicle. 
     
     
         9 . The method of  claim 8 , wherein the point on the vehicle is a two-dimensional point. 
     
     
         10 . The method of  claim 8 , wherein the point on the vehicle is a point on the LIDAR sensor. 
     
     
         11 . The method of  claim 8 , wherein removing data points of the first model further includes identifying a pair of data points from the modified convex hull using the point on the vehicle. 
     
     
         12 . The method of  claim 11 , wherein the pair of data points correspond to two data points in the modified convex hull on either side of the point on the vehicle. 
     
     
         13 . The method of  claim 11 , wherein the pair of data points correspond to two data points in the modified convex hull that are closest to the point on the vehicle. 
     
     
         14 . The method of  claim 13 , wherein removing data points of the first model includes identifying data points of the modified convex hull or the first model that are located farther from the point on the vehicle than either of the pair of data points. 
     
     
         15 . The method of  claim 1 , wherein removing data points of the first model includes identifying a pair of data points from the two-dimensional data points using a reference vector originating at a point on the vehicle. 
     
     
         16 . The method of  claim 15 , wherein the point on the vehicle is a two-dimensional point. 
     
     
         17 . The method of  claim 15 , wherein the reference vector extends in a direction of one of a heading of the vehicle or an orientation of the vehicle. 
     
     
         18 . The method of  claim 15 , wherein the reference vector extends in a direction of a direction of traffic of a lane in which the vehicle is traveling. 
     
     
         19 . The method of  claim 15 , wherein a first of the pair of data points is identified as a data point of the two-dimensional data points that is positioned at a smallest angle relative to the reference angle of the vehicle, and a second of the pair of data points is identified as a data point of the two-dimensional data points that is positioned at a greatest angle relative to the reference angle of the vehicle. 
     
     
         20 . The method of  claim 1 , wherein controlling the vehicle based on the second model includes passing data of the second model to one or more systems of the vehicle in order to make driving decisions for the vehicle using the second model.