Convolution operator selection
A vehicle system includes one or more sensors configured to capture aspects of an environment and a computing device. The computing device is configured to receive information about the environment captured by the one or more sensors, determine one or more structures within the environment based on the received information, select a kernel that is parameterized for predicting a vehicle trajectory based on the one or more structures determined within the environment, and perform a convolution of the selected kernel and an array defining the environment, wherein the convolution predicts a future trajectory of a vehicle within the environment.
1. A vehicle system comprising:
one or more sensors configured to capture aspects of an environment; and
a computing device configured to:
receive information about the environment captured by the one or more sensors,
determine one or more structures within the environment based on the received information,
select a kernel from a plurality of kernels that correspond to various environments, the selected kernel is parameterized for predicting a vehicle trajectory based on the one or more structures determined within the environment,
perform a convolution of the selected kernel and an array defining the environment, wherein the convolution predicts a future trajectory of a vehicle within the environment, wherein the future trajectory of the vehicle defines one or more predicted future locations of the vehicle in the environment,
obtain a sensed vehicle trajectory of the vehicle,
compare the sensed vehicle trajectory of the vehicle with the predicted future trajectory of the vehicle, and
select, in response to the comparison indicating an inaccurate predicted future trajectory of the vehicle, a second kernel that is parameterized for predicting a vehicle trajectory, wherein the second kernel is based on the one or more structures determined within the environment and the comparison of the sensed vehicle trajectory of the vehicle with the predicted future trajectory of the vehicle.
2. The vehicle system of claim 1 , wherein the computing device is further configured to generate one or more kernels comprising parameters that are learned based on the one or more structures of the environment and vehicle behaviors corresponding to the environment.
3. The vehicle system of claim 2 , wherein the vehicle behaviors comprise at least one of expected vehicle trajectories within the environment, a vehicle speed, or a rule of a road that applies to the environment.
4. The vehicle system of claim 1 , wherein the computing device is further configured to define a plurality of discretized locations within the environment and generate an occupancy map based on the predicted future trajectory of the vehicle, wherein the occupancy map is defined by the plurality of discretized locations within the environment defining a probability distribution of a likelihood that the vehicle will be located at the plurality of discretized locations within the environment.
5. The vehicle system of claim 4 , wherein the computing device is further configured to provide the probability distribution to an advanced driver assistance system for determining an action to take in response to the probability distribution.
6. The vehicle system of claim 1 , wherein the one or more structures comprises identification of at least one of a width of a road, a number of lanes of the road, a weather condition, a time of day, a speed limit, a traffic density, or an urban or rural environment.
7. The vehicle system of claim 1 , wherein the one or more structures comprises the vehicle.
8. The vehicle system of claim 1 , wherein the kernel is based on a heat equation.
9. A computing device configured to:
receive information about an environment;
determine one or more structures within the environment based on the received information;
select a kernel from a plurality of kernels that correspond to various environments, the selected kernel is parameterized for predicting a vehicle trajectory based on the one or more structures determined within the environment;
perform a convolution of the selected kernel and an array defining the environment, wherein the convolution predicts a future trajectory of a vehicle within the environment, wherein the future trajectory of the vehicle defines one or more predicted future locations of the vehicle in the environment;
obtain a sensed vehicle trajectory of the vehicle;
compare the sensed vehicle trajectory of the vehicle with the predicted future trajectory of the vehicle; and
select, in response to the comparison indicating an inaccurate predicted future trajectory of the vehicle, a second kernel that is parameterized for predicting a vehicle trajectory, wherein the second kernel is based on the one or more structures determined within the environment and the comparison of the sensed vehicle trajectory of the vehicle with the predicted future trajectory of the vehicle.
10. The computing device of claim 9 , wherein the computing device is further configured to generate one or more kernels comprising parameters that are learned based on the one or more structures of the environment and vehicle behaviors corresponding to the environment.
11. The computing device of claim 10 , wherein the vehicle behaviors comprise at least one of expected vehicle trajectories within the environment, a vehicle speed, or a rule of a road that applies to the environment.
12. The computing device of claim 9 , wherein the computing device is further configured to define a plurality of discretized locations within the environment and generate an occupancy map based on the predicted future trajectory of the vehicle, wherein the occupancy map is defined by the plurality of discretized locations within the environment defining a probability distribution of a likelihood that the vehicle will be located at the plurality of discretized locations within the environment.
13. The computing device of claim 12 , wherein the computing device is further configured to provide the probability distribution to an advanced driver assistance system for determining an action to take in response to the probability distribution.
14. The computing device of claim 9 , wherein the one or more structures comprises identification of at least one of a width of a road, a number of lanes of the road, a weather condition, a time of day, a speed limit, a traffic density, or an urban or rural environment.
15. A method for predicting future events, the method comprising:
receiving information about an environment;
determining one or more structures within the environment based on the received information;
selecting a kernel from a plurality of kernels that correspond to various environments, the selected kernel is parameterized for predicting a vehicle trajectory based on the one or more structures determined within the environment;
performing a convolution of the selected kernel and an array defining the environment, wherein the convolution predicts a future trajectory of a vehicle within the environment, wherein the future trajectory of the vehicle defines one or more predicted future locations of the vehicle in the environment;
obtaining a sensed vehicle trajectory of the vehicle;
comparing the sensed vehicle trajectory of the vehicle with the predicted future trajectory of the vehicle; and
selecting, in response to the comparison indicating an inaccurate predicted future trajectory of the vehicle, a second kernel that is parameterized for predicting a vehicle trajectory, wherein the second kernel is based on the one or more structures determined within the environment and the comparison of the sensed vehicle trajectory of the vehicle with the predicted future trajectory of the vehicle.
16. The method of claim 15 , further comprising:
defining a plurality of discretized locations within the environment; and
generating an occupancy map based on the predicted future trajectory of the vehicle, wherein the occupancy map is defined by the plurality of discretized locations within the environment defining a probability distribution of a likelihood that the vehicle will be located at the plurality of discretized locations within the environment.
17. The method of claim 16 , further comprising:
providing the probability distribution to an advanced driver assistance system for determining an action to take in response to the probability distribution.