Autonomous vehicles featuring vehicle intention system
The present disclosure provides autonomous vehicles that include a vehicle intention system that provides intention signals indicative of an intention of the autonomous vehicle. In particular, in one example, the vehicle intention system can obtain one or more operational messages from various systems or components of an autonomous vehicle that operate to control the autonomous vehicle. The operational messages can include operational data regarding the control or operation of the autonomous vehicle. The vehicle intention system can determine an intention of the autonomous vehicle based at least in part on the one or more operational messages. The vehicle intention system can output one or more intention signals that indicate the determined intention of the autonomous vehicle. For example, the vehicle intention system can publish intention messages that indicate the determined intention to one or more components or systems that consume the intention messages.
1. A computing system of an autonomous vehicle, the computing system comprising:
multiple different systems or subsystems that contribute to operation of the autonomous vehicle, wherein the multiple different systems or subsystems generate one or more operational messages that comprise operational data indicative of operation of the autonomous vehicle; and
a vehicle intention system that is separate from and operates independently of the multiple different systems or subsystems, the vehicle intention system comprising:
one or more non-transitory computer-readable media that store instructions that, when executed by one or more processors, cause the one or more processors to perform operations, the operations comprising:
obtaining the one or more operational messages from the multiple different systems or subsystems that contribute to operation of the autonomous vehicle, the one or more operational messages comprising the operational data indicative of operation of the autonomous vehicle;
determining an intention of the autonomous vehicle based at least in part on the one or more operational messages, wherein determining the intention of the autonomous vehicle based at least in part on the one or more operational messages comprises identifying a particular object that is a cause of the intention of the autonomous vehicle, and wherein identifying the particular object that is a cause of the intention of the autonomous vehicle comprises:
determining, based on one or more cost functions, a total cost associated with a surrounding environment of the autonomous vehicle, the surrounding environment including a plurality of objects;
iteratively removing a respective one of the plurality of objects to determine a respective reduction in the total cost as a result of removing such object from the surrounding environment; and
identifying the object for which removal results in the largest reduction in the total cost as the particular object that is the cause of the intention of the autonomous vehicle; and
outputting an intention signal that indicates the intention of the autonomous vehicle to one or more consumers, wherein outputting the intention signal comprises outputting the intention signal that identifies the particular object as the cause of the intention of the autonomous vehicle.
2. The computing system of claim 1 , wherein the multiple different systems or subsystems that contribute to operation of the autonomous vehicle comprise an autonomy computing system that controls motion of the autonomous vehicle based at least in part on sensor data, and wherein the one or more consumers comprise one or more of the multiple different systems or subsystems that are included in the autonomy computing system.
3. The computing system of claim 2 , wherein the one or more of the multiple different systems or subsystems that are included in the autonomy computing system that consume the intention signal comprise one or more of: a perception system, a prediction system, and a motion planning system.
4. The computing system of claim 1 , wherein the one or more consumers comprise a human machine interface system that is included within the autonomous vehicle and that indicates the intention of the autonomous vehicle to one or more humans physically located within the autonomous vehicle.
5. The computing system of claim 1 , wherein the one or more consumers comprise one or more additional autonomous vehicles.
6. The computing system of claim 1 , wherein:
the multiple different systems or subsystems that contribute to operation of the autonomous vehicle comprise an autonomy computing system that controls motion of the autonomous vehicle by processing sensor data descriptive of the plurality of objects in the surrounding environment of the autonomous vehicle;
in response to the intention signal that identifies the particular object, the autonomy computing system adjusts its operations to allocate additional processing resources to the particular object that is the cause of the intention of the autonomous vehicle.
7. The computing system of claim 1 , wherein the total cost associated with the surrounding environment comprises a sum of a cost field that describes a respective cost associated with each location in the surrounding environment.
8. The computing system of claim 1 , wherein execution of the instructions by the one or more processors further cause the one or more processors to perform additional operations, the additional operations comprising:
logging the intention signals output by the vehicle intention system over time; and
using the logged intention signals as training data to train one or more machine-learned models.
9. The computing system of claim 1 , wherein:
the multiple different systems or subsystems that contribute to operation of the autonomous vehicle comprise an autonomy computing system that controls motion of the autonomous vehicle by processing raw sensor data;
the multiple different systems or subsystems that contribute to operation of the autonomous vehicle comprise a reflex control system that interrupts control of the motion of the autonomous vehicle by the autonomy computing system when the raw sensor data indicates an imminent collision;
obtaining the one or more operational messages comprises obtaining the one or more operational messages from both the autonomy computing system and the reflex control system; and
determining the intention of the autonomous vehicle comprises determining the intention of the autonomous vehicle by synthesizing the one or more operational messages received from both the autonomy computing system and the reflex control system.
10. The computing system of claim 1 , wherein:
the computing system further comprises a machine-learned intention model configured to receive the operational data and, in response, provide the intention of the autonomous vehicle; and
determining the intention of the autonomous vehicle comprises:
inputting the operational data included in the one or more operational messages into the machine-learned intention model; and
receiving the intention of the autonomous vehicle as an output of the machine-learned intention model.
11. The computing system of claim 1 , wherein:
the computing system further comprises a machine-learned intention model configured to receive the operational data and, in response, identify a particular object that is a cause of the intention of the autonomous vehicle; and
determining the intention of the autonomous vehicle comprises:
inputting the operational data included in the one or more operational messages into the machine-learned intention model; and
receiving identification of the particular object that is the cause of the intention of the autonomous vehicle as an output of the machine-learned intention model.
12. The computing system of claim 1 , wherein determining the intention of the autonomous vehicle comprises applying a set of rules to the operational data to determine the intention of the autonomous vehicle.
13. The computing system of claim 1 , wherein:
each of the one or more consumers has subscribed to a respective subset of particular intentions; and
outputting the intention signal that indicates the intention of the autonomous vehicle to the one or more consumers comprises outputting the intention signal to only those consumers that have subscribed to the particular intention indicated by the intention signal.
14. An autonomous vehicle, comprising:
multiple different systems or subsystems that contribute to operation of the autonomous vehicle, wherein the multiple different systems or subsystems generate one or more operational messages that comprise operational data indicative of operation of the autonomous vehicle; and
a vehicle intention system that is separate from and operates independently of the multiple different systems or subsystems, the vehicle intention system comprising:
one or more non-transitory computer-readable media that store instructions that, when executed by one or more processors, cause the one or more processors to implement a vehicle intention system that performs operations, the operations comprising:
obtaining the one or more operational messages from one or more producers, the one or more operational messages comprising operational data indicative of operation of the autonomous vehicle;
determining an intention of the autonomous vehicle based at least in part on the one or more operational messages, wherein determining the intention of the autonomous vehicle based at least in part on the one or more operational messages comprises identifying a particular object that is a cause of the intention of the autonomous vehicle, and wherein identifying the particular object that is a cause of the intention of the autonomous vehicle comprises:
determining, based on one or more cost functions, a total cost associated with a surrounding environment of the autonomous vehicle, the surrounding environment including a plurality of objects;
iteratively removing a respective one of the plurality of objects to determine a respective reduction in the total cost as a result of removing such object from the surrounding environment; and
identifying the object for which removal results in the largest reduction in the total cost as the particular object that is the cause of the intention of the autonomous vehicle; and
outputting an intention signal that indicates the intention of the autonomous vehicle to one or more consumers, wherein outputting the intention signal comprises outputting the intention signal that identifies the particular object as the cause of the intention of the autonomous vehicle.
15. The autonomous vehicle of claim 14 , wherein the one or more consumers comprise one or more of:
an autonomy computing system that controls motion of the autonomous vehicle based at least in part on sensor data;
a human machine interface system that is included within the autonomous vehicle and that indicates the intention of the autonomous vehicle to one or more humans physically located within the autonomous vehicle; and
one or more additional autonomous vehicles.
16. The autonomous vehicle of claim 14 , wherein:
the multiple different systems or subsystems that contribute to operation of the autonomous vehicle comprise an autonomy computing system that controls motion of the autonomous vehicle by processing sensor data descriptive of the plurality of objects in the surrounding environment of the autonomous vehicle;
in response to the intention signal that identifies the particular object, the autonomy computing system adjusts its operations to allocate additional processing resources to the particular object that is the cause of the intention of the autonomous vehicle.
17. A computing system of an autonomous vehicle, the computing system comprising:
one or more processors; and
one or more non-transitory computer-readable media that store instructions that, when executed by the one or more processors, cause the one or more processors to implement a vehicle intention system that performs operations, the operations comprising:
obtaining one or more operational messages from one or more vehicle systems that contribute to operation of the autonomous vehicle, the one or more operational messages comprising operational data indicative of operation of the autonomous vehicle;
determining an intention of the autonomous vehicle based at least in part on the one or more operational messages, wherein determining the intention of the autonomous vehicle based at least in part on the one or more operational messages comprises identifying a particular object that is a cause of the intention of the autonomous vehicle, wherein identifying the particular object that is the cause of the intention of the autonomous vehicle comprises:
determining, based on one or more cost functions, a total cost associated with a surrounding environment of the autonomous vehicle, the surrounding environment including a plurality of objects;
iteratively removing a respective one of the plurality of objects to determine a respective reduction in the total cost as a result of removing such object from the surrounding environment; and
identifying the object for which removal results in the largest reduction in the total cost as the particular object that is the cause of the intention of the autonomous vehicle; and
outputting an intention signal that indicates the intention of the autonomous vehicle to one or more consumers, wherein outputting the intention signal comprises outputting the intention signal that identifies the particular object as the cause of the intention of the autonomous vehicle.