Behavior planning for autonomous vehicles in yield scenarios
In various examples, a yield scenario may be identified for a first vehicle. A wait element is received that encodes a first path for the first vehicle to traverse a yield area and a second path for a second vehicle to traverse the yield area. The first path is employed to determine a first trajectory in the yield area for the first vehicle based at least on a first location of the first vehicle at a time and the second path is employed to determine a second trajectory in the yield area for the second vehicle based at least on a second location of the second vehicle at the time. To operate the first vehicle in accordance with a wait state, it may be determined whether there is a conflict between the first trajectory and the second trajectory, where the wait state defines a yielding behavior for the first vehicle.
1 . A system comprising:
one or more processors to perform operations including:
determining, based at least on sensor data obtained using at least one sensor of a machine, a path for the machine to traverse an area in an environment;
applying, to the path, at least two categories of yield behavior that encode one or more traffic rules using a sequence of steps for controlling the machine with respect to one or more contention points of the area to comply with the one or more traffic rules, a first category of the at least two categories having a first step including a yield maneuver, and one or more first conditions for transitioning to a subsequent yield maneuver in second step of a second category of the at least two categories in the sequence, the second step including one or more second conditions for terminating the sequence of steps based at least on clearing the one or more contention points;
based at least on the applying and detecting an entrance to the area along the path, controlling the machine to traverse the area using the path and the at least two categories of yield behavior, the controlling including tracking, using the one or more first conditions and the one or more second conditions, progress of the machine through the steps of the sequence to advance the machine along the path through the area, and ending the use of the sequence of steps for the controlling based at least on determining the one or more second conditions are satisfied.
2 . The system of claim 1 , wherein the one or more first conditions correspond to at least one of:
a speed of the machine;
a distance of the machine to at least one contention point of the one or more contention points; or
a wait time for the machine.
3 . The system of claim 1 , wherein the first category of yield behavior includes a stop with respect to at least one contention point of the one or more contention points until the one or more first conditions are satisfied.
4 . The system of claim 1 , wherein the first category of yield behavior is applied to a first segment of the path to generate a first portion of a trajectory that corresponds to the first category of yield behavior, and the operations further include applying, to a second segment of the path, the second category of yield behavior to generate a second portion of the trajectory that corresponds to the second category of yield behavior.
5 . The system of claim 1 , wherein the controlling initiates the use of the sequence of steps for the controlling based at least on the detecting of the entrance to the area along the path, and the ending the use of the sequence of steps for the controlling is based at least on detecting an exit to the area along the path.
6 . The system of claim 1 , wherein the applying is based at least on selecting the at least two categories of yield behavior as a subset of primitive states based at least on matching the area to the subset of primitive states.
7 . The system of claim 1 , wherein the operations further include:
localizing the machine to one or more locations in a map of the environment, wherein the map encodes the one or more traffic rules that correspond to the one or more locations;
detecting, using perception components of the machine, one or more dynamic traffic signal states in the environment; and
based at least on the localizing, selecting, for the applying, the first category of yield behavior from a plurality of categories of yield behavior based at least on the first category of yield behavior corresponding to the one or more traffic rules and the one or more dynamic traffic signal states.
8 . The system of claim 1 , wherein the sequence of steps is represented as a state machine in which each step of the steps corresponds to a respective state of the state machine.
9 . The system of claim 1 , wherein the system is comprised in at least one of:
a control computing system for an autonomous or semi-autonomous machine;
a perception computing system for an autonomous or semi-autonomous machine;
a system for performing simulation operations;
a system for performing deep learning operations;
a system implemented using an edge computing device;
a system implemented using a robot;
a system incorporating one or more virtual machines (VMs);
a system implemented at least partially in a data center; or
a system implemented at least partially using cloud computing resources.
10 . A method comprising:
selecting, for a path and based at least on sensor data obtained using at least one sensor of a machine, at least two categories of yield behavior that encode one or more traffic rules using a sequence of steps for controlling the machine with respect to one or more contention points of an area in an environment to comply with the one or more traffic rules, a first category of the at least two categories having a first step including a yield maneuver, and one or more first conditions for transitioning to a subsequent yield maneuver in a second step of a second category of the at least two categories in the sequence, the second step including one or more second conditions for terminating the sequence of steps based at least on clearing the one or more contention points;
based at least on the selecting and detecting an entrance to the area along the path, controlling the machine to traverse the area using the path and the at least two categories of yield behavior, the controlling including tracking, using the one or more first conditions and the one or more second conditions, progress of the machine through the steps of the sequence to advance the machine along the path through the area, and ending the use of the sequence of steps for the controlling based at least on determining the one or more second conditions are satisfied.
11 . The method of claim 10 , wherein the detecting of the entrance along the path signals a beginning of the use of the sequence of the steps for the controlling of the machine.
12 . The method of claim 10 , wherein the one or more first conditions correspond to at least one of a wait time for the machine, a speed of the machine, or a distance of the machine to at least one contention point of the one or more contention points.
13 . The method of claim 10 , wherein the controlling prevents the machine from advancing along the path until jointly determining that each step in the sequence of steps is capable of being cleared without collision.
14 . The method of claim 10 , wherein the at least two categories of yield behavior include a stop at entry category and a yield contention point category.
15 . The method of claim 10 , comprising:
localizing, using the sensor data, the machine to one or more locations in a map of the environment, wherein the map encodes the one or more traffic rules that correspond to the one or more locations; and
detecting, using perception components of the machine, one or more dynamic traffic signal states in the environment,
wherein the selecting of the at least two categories of yield behavior is from a plurality of categories of yield behavior based at least on the at least two categories of yield behavior corresponding to the one or more traffic rules and the one or more dynamic traffic signal states.
16 . At least one processor comprising:
one or more circuits to, based at least on detecting an entrance to an area in an environment along a path, control a machine to traverse the area using the path and at least two categories of yield behavior assigned to the path,
the at least two categories of yield behavior encoding one or more traffic rules using a sequence of steps for controlling the machine with respect to one or more contention points of the area to comply with the one or more traffic rules, a first category of the at least two categories having a first step including a yield maneuver, and one or more first conditions for transitioning to a subsequent yield maneuver in a second step of a second category of the at least two categories in the sequence, the second step including one or more second conditions for terminating the sequence of steps based at least on clearing the one or more contention points, and
the control including tracking, using the one or more first conditions and the one or more second conditions, progress of the machine through the steps of the sequence to advance the machine along the path through the area, and ending the use of the sequence of steps for the control based at least on determining the one or more second conditions are satisfied.
17 . The at least one processor of claim 16 , wherein the second category of yield behavior is defined with respect to:
one or more crossing points between the path for the machine and one or more paths for one or more agents; or
one or more merge points between the path for the machine and the one or more paths for the one or more agents.
18 . The at least one processor of claim 16 , wherein the second category of yield behavior includes a yield with respect to at least one contention point between the machine and one or more agents until the one or more second conditions are satisfied.
19 . The at least one processor of claim 16 , wherein the control uses a trajectory including a first portion corresponding to the first category of yield behavior, and a second portion corresponding to the second category of yield behavior.
20 . The at least one processor of claim 16 , wherein the at least one processor is comprised in at least one of:
a control computing system for an autonomous or semi-autonomous machine;
a perception computing system for an autonomous or semi-autonomous machine;
a system for performing simulation operations;
a system for performing deep learning operations;
a system implemented using an edge computing device;
a system implemented using a robot;
a system incorporating one or more virtual machines (VMs);
a system implemented at least partially in a data center; or
a system implemented at least partially using cloud computing resources.