Planning accommodations for reversing vehicles
Techniques for determining that a first vehicle is associated with a reverse state, and controlling a second vehicle based on the reverse state, are described herein. In some examples, the first vehicle may provide an indication that the first vehicle will be executing a reverse maneuver, such as with reverse lights on the vehicle or by positioning at an angle relative to a road or parking space to allow for the reverse maneuver into a desired location. A planning system of the second vehicle (such as an autonomous vehicle) may receive sensor data and determine a variety of these indications to determine a probability that the vehicle is going to execute a reverse maneuver. The second vehicle can further determine a likely trajectory of the reverse maneuver and can provide appropriate accommodations (e.g., time and/or space) to allow the second vehicle to execute the maneuver safely and efficiently.
1. A system comprising:
one or more processors; and
one or more non-transitory computer-readable media storing instructions executable by the one or more processors, wherein the instructions, when executed, cause the system to perform operations comprising:
obtaining image data of an environment using an image sensor associated with the system;
determining, based at least in part on the image data, a first feature indicative of a vehicle reverse light state of a vehicle proximate the system, the vehicle having a location within the environment;
determining, based at least in part on the image data, a velocity of the vehicle;
determining, based at least in part on map data and the location, a speed limit associated with the location;
determining a difference between the velocity of the vehicle and the speed limit associated with the location;
predicting a vehicle trajectory, based at least in part on inputting the difference between the velocity of the vehicle and the speed limit associated with the location, the first feature and a second feature associated with the map data, into a machine-learned model, wherein predicting the vehicle trajectory includes extending a width of the vehicle trajectory based on a width of the vehicle; and
determining, based at least in part on the vehicle trajectory, a second trajectory for the system to traverse the environment;
wherein the second trajectory avoids the vehicle trajectory.
2. The system of claim 1 , wherein the instructions, when executed, cause the system to perform operations further comprising:
determining the width of the vehicle; and
associating the width of the vehicle with a length of the vehicle trajectory,
wherein the second trajectory is configured to cause the system to avoid an area defined by the width of the vehicle trajectory along the length of the vehicle trajectory.
3. The system of claim 1 , wherein the vehicle trajectory is associated with the vehicle preparing to parallel park and the second trajectory is configured to cause the system to one or more of wait for the vehicle to finish parking or enter an oncoming traffic lane and travel along the second trajectory.
4. The system of claim 1 , wherein the vehicle is a first vehicle and the instructions, when executed, cause the system to perform operations further comprising:
determining a second difference between a first velocity of the first vehicle and a second velocity of a second vehicle in the environment,
wherein predicting the vehicle trajectory is further based at least in part on inputting the second difference into the machine-learned model or another machine-learned model.
5. The system of claim 1 , wherein the instructions, when executed, cause the system to perform operations further comprising:
determining, based at least in part on the image data, a third feature being indicative of a heading of the vehicle relative to a lane,
wherein predicting the vehicle trajectory is further based at least in part on inputting the third feature into the machine-learned model or another machine-learned model.
6. The system of claim 1 , wherein the machine-learned model is trained using training image data to indicate:
a ground truth of the vehicle reverse light state at a time that the training image data is captured, and
a heading of a training vehicle depicted in the training image data for an amount of time proximate the time that the training image data was captured.
7. The system of claim 1 , wherein the second feature associated with the map data is one of a current lane type associated with a current lane occupied by the vehicle, a neighbor lane type associated with a neighbor lane proximate the current lane occupied by the vehicle, or an indication of a junction proximate the vehicle.
8. The system of claim 7 , wherein predicting the vehicle trajectory is further based at least in part on the current lane type.
9. The system of claim 1 , wherein predicting the vehicle trajectory further comprises predicting that the vehicle will proceed in reverse at the velocity of the vehicle for a predetermined amount of time.
10. A method comprising:
receiving sensor data captured by a sensor associated with an autonomous vehicle operating in an environment;
determining a feature associated with a vehicle proximate the autonomous vehicle, the feature being indicative of a reverse light state of the vehicle;
determining, based at least in part on the sensor data, a velocity of the vehicle;
determining, based at least in part on map data and a location of the autonomous vehicle within the environment, a speed limit associated with the location;
determining a difference between the velocity of the vehicle and the speed limit associated with the location;
inputting the feature and the difference between the velocity of the vehicle and the speed limit associated with the location into a machine-learned model trained to predict vehicle behavior based at least in part on the feature;
predicting, based at least in part on an indication that the vehicle is associated with the reverse light state and a second feature of the environment determined from the map data, a vehicle trajectory, wherein predicting the vehicle trajectory includes extending a width of the vehicle trajectory based on a width of the vehicle; and
determining, based at least in part on the vehicle trajectory, a control trajectory for the autonomous vehicle to traverse the environment such that the autonomous vehicle avoids the vehicle trajectory.
11. The method of claim 10 , further comprising:
determining the width of the vehicle;
associating the width of the vehicle with a length of the vehicle trajectory; and
controlling the autonomous vehicle to travel along the control trajectory that avoids an area defined by the width of the vehicle trajectory along the length of the vehicle trajectory.
12. The method of claim 10 , wherein the vehicle trajectory is indicative of the vehicle preparing to parallel park, the method further comprising:
controlling the autonomous vehicle to one or more of:
wait for the vehicle to complete parking, or
enter an oncoming traffic lane and travel along the control trajectory.
13. The method of claim 10 , further comprising:
maintaining a position of the autonomous vehicle for a duration of time associated with the vehicle executing the vehicle trajectory.
14. The method of claim 10 , further comprising:
determining a region proximate the vehicle that the vehicle is permitted to reverse into,
wherein predicting the vehicle trajectory is further based on the region proximate the vehicle.
15. The method of claim 14 , wherein the region comprises one or more of:
a lane of traffic,
a parking space, or
a driveway.
16. One or more non-transitory computer-readable media storing instructions that when executed by one or more processors perform operations comprising:
receiving sensor data captured by a sensor associated with an autonomous vehicle operating in an environment;
determining, based at least in part on the sensor data, a feature indicative of a reverse light state of a vehicle;
determining, based at least in part on the sensor data, a velocity of the vehicle;
determining, based at least in part on map data and a location of the autonomous vehicle within the environment, a speed limit associated with the location;
determining a difference between the velocity of the vehicle and the speed limit associated with the location;
predicting a first trajectory of the vehicle, based at least in part on inputting the difference between the velocity of the vehicle and the speed limit associated with the location, the feature, and a second feature of the environment determined from the map data into a machine-learned model, wherein predicting the first trajectory includes extending a width of the first trajectory based on a width of the vehicle; and
determining, based at least in part on the first trajectory, a second trajectory for the autonomous vehicle to traverse the environment such that the autonomous vehicle circumnavigates the first trajectory.
17. The one or more non-transitory computer-readable media of claim 16 , wherein determining the feature comprises:
inputting at least a portion of the sensor data associated with the vehicle into the machine-learned model; and
receiving, from the machine-learned model, the feature.