Autonomous vehicle motion control systems and methods
Systems and methods for controlling the motion of an autonomous are provided. In one example embodiment, a computer-implemented method includes obtaining data associated with an object within a surrounding environment of an autonomous vehicle. The data associated with the object is indicative of a predicted motion trajectory of the object. The method includes determining a vehicle action sequence based at least in part on the predicted motion trajectory of the object. The vehicle action sequence is indicative of a plurality of vehicle actions for the autonomous vehicle at a plurality of respective time steps associated with the predicted motion trajectory. The method includes determining a motion plan for the autonomous vehicle based at least in part on the vehicle action sequence. The method includes causing the autonomous vehicle to initiate motion control in accordance with at least a portion of the motion plan.
1 . A computer-implemented method comprising:
generating training data comprising driving log data and one or more first labels for the driving log data, the driving log data associated with a training object in an environment of a vehicle, the first labels respectively indicative of whether the training object is blocking a travel way at respective time steps and at respective locations associated with a trajectory of the vehicle; and
training a machine-learned model using the training data, wherein the trained machine-learned model is configured to be executed by an autonomous vehicle operating within the environment.
2 . The computer-implemented method of claim 1 , wherein the driving log data is based on sensor data acquired by one or more sensors located on the vehicle as the vehicle travels on the travel way.
3 . The computer-implemented method of claim 1 , further comprising: providing a portion of the training data as input to the machine-learned model;
determining, based on an output of the machine-learned model, in response to receipt of the portion of the training data provided as input, and relative to the first labels utilized as ground-truth data, an accuracy level of the machine-learned model; and
updating the machine-learned model based on the accuracy level.
4 . The computer-implemented method of claim 3 , wherein updating the machine-learned model based on the accuracy level comprises employing backwards propagation of errors.
5 . The computer-implemented method of claim 1 , wherein the machine-learned model is configured to make a blocking decision for a particular time step based at least in part on a blocking decision determined for one or more previous time steps.
6 . The computer-implemented method of claim 1 , wherein the first labels for the driving log data are human-labeled.
7 . The computer-implemented method of claim 1 , wherein the first labels for the driving log data are machine-labeled.
8 . The computer-implemented method of claim 1 , wherein the training object comprises a pedestrian or another vehicle.
9 . The computer-implemented method of claim 1 , wherein the training data comprises one or more second labels for the driving log data, the second labels for the driving log data comprising a vehicle action at the respective time steps.
10 . The computer-implemented method of claim 9 , wherein the vehicle action at the respective time steps is determined as one of a pass action or a queue action.
11 . The computer-implemented method of claim 9 , further comprising:
training a machine-learned vehicle action model using the training data including the one or more second labels, wherein the trained machine-learned vehicle action model is configured to be executed by the autonomous vehicle operating within the environment.
12 . The computer-implemented method of claim 11 , wherein the machine-learned vehicle action model is configured to determine a vehicle action sequence comprising respective discrete vehicle actions with respect to an object in the environment of the autonomous vehicle.
13 . A computing system comprising:
one or more processors; and
one or more non-transitory computer-readable media that store instructions for execution by the one or more processors to cause the one or more processors to perform operations comprising:
generating training data comprising driving log data and one or more first labels for the driving log data, the driving log data associated with a training object in an environment of a vehicle, the first labels respectively indicative of whether the training object is blocking a travel way at respective time steps and at respective locations associated with a trajectory of the vehicle; and
training a machine-learned model using the training data, wherein the trained machine-learned model is configured to be executed by an autonomous vehicle operating within the environment.
14 . The computing system of claim 13 , wherein the driving log data is based on sensor data acquired by one or more sensors located on the vehicle as the vehicle travels on the travel way.
15 . The computing system of claim 13 , the operations further comprising: providing a portion of the training data as input to the machine-learned model;
determining, based on an output of the machine-learned model, in response to receipt of the portion of the training data provided as input, and relative to the first labels utilized as ground-truth data, an accuracy level of the machine-learned model; and
updating the machine-learned model based on the accuracy level.
16 . The computing system of claim 13 , wherein the training data comprises one or more second labels for the driving log data, the second labels for the driving log data comprising a vehicle action at the respective time steps.
17 . The computing system of claim 16 , the operations further comprising:
training a machine-learned vehicle action model using the training data including the one or more second labels, wherein the trained machine-learned vehicle action model is configured to be executed by the autonomous vehicle operating within the environment.
18 . The computing system of claim 17 , wherein the machine-learned vehicle action model is configured to determine a vehicle action sequence comprising respective discrete vehicle actions with respect to an object in the environment of the autonomous vehicle.
19 . One or more non-transitory computer-readable media that store instructions for execution by one or more processors to cause the one or more processors to perform operations comprising:
generating training data comprising driving log data and one or more first labels for the driving log data, the driving log data associated with a training object in an environment of a vehicle, the first labels respectively indicative of whether the training object is blocking a travel way at respective time steps and at respective locations associated with a trajectory of the vehicle; and
training a machine-learned model using the training data, wherein the trained machine-learned model is configured to be executed by an autonomous vehicle operating within the environment.
20 . The one or more non-transitory computer-readable media of claim 19 , wherein: the training data comprises one or more second labels for the driving log data, the second labels for the driving log data comprising a vehicle action at the respective time steps; and
the operations further comprise training a machine-learned vehicle action model using the training data including the one or more second labels, wherein the trained machine-learned vehicle action model is configured to be executed by the autonomous vehicle operating within the environment.