Determining the pick-up/drop-off state of a mass transit vehicle
Systems and methods for determining a passenger pick-up/drop-off (PUDO) state of a mass transit vehicle by an autonomous vehicle (AV) are disclosed. A system includes a memory one or more processing devices, coupled to the memory, configured to perform operations that include identifying a mass transit vehicle in an environment of an AV, responsive to determining the mass transit vehicle is in a PUDO state, providing movement instructions to a planning system of the AV, and autonomously modifying operation of the AV based on the movement instructions.
1 . A system, comprising:
a memory; and
one or more processing devices, coupled to the memory, configured to perform operations comprising:
identifying a mass transit vehicle in an environment of an autonomous vehicle (AV);
responsive to determining the mass transit vehicle is in a pick-up/drop-off (PUDO) state and that there is a lack of pedestrian activity involving boarding or disembarking from the mass transit vehicle for a threshold time interval, providing movement instructions to a planning system of the AV to at least partially circumnavigate the mass transit vehicle; and
modifying operation of the AV based on the movement instructions.
2 . The system of claim 1 , wherein identifying the mass transit vehicle comprises:
obtaining, based on sensor data from a sensing system of the AV, a distance between a pair of rails on which the mass transit vehicle is positioned; and
identifying the mass transit vehicle as a cable car based on the distance between the pair of rails and a reference distance.
3 . The system of claim 1 , wherein identifying the mass transit vehicle comprises identifying a cable car based on map data from a mapping system of the AV, the map data comprising a plurality of cable car routes.
4 . The system of claim 1 , wherein determining the mass transit vehicle is in the PUDO state comprises determining, based on map data, at least one of:
the mass transit vehicle is within a first threshold distance from a passenger PUDO zone; or
the mass transit vehicle is entering an intersection, and the passenger PUDO zone is within a second threshold distance from the intersection.
5 . The system of claim 1 , wherein determining the mass transit vehicle is in the PUDO state comprises:
obtaining sensor data from a sensing system of the AV, wherein the sensor data comprises data characterizing a gaze of a passenger of the mass transit vehicle; and
applying a machine learning model (MLM) onboard the AV to the sensor data to generate an output indicating that the gaze of the passenger indicates that the mass transit vehicle is in the PUDO state.
6 . The system of claim 1 , wherein determining the mass transit vehicle is in the PUDO state comprises:
obtaining sensor data from a sensing system of the AV, wherein the sensor data comprises data characterizing a gaze of a pedestrian within a threshold distance from the mass transit vehicle; and
applying a MLM onboard the AV to the sensor data to generate an output indicating that the gaze of the pedestrian indicates that the mass transit vehicle is in the PUDO state.
7 . The system of claim 1 , wherein determining the mass transit vehicle is in the PUDO state comprises:
obtaining sensor data from a sensing system of the AV, wherein the sensor data comprises data characterizing a trajectory of a pedestrian within a threshold distance from the mass transit vehicle; and
applying a MLM onboard the AV to the sensor data to generate an output indicating that the trajectory of the pedestrian indicates the mass transit vehicle is in the PUDO state.
8 . The system of claim 1 , wherein determining the mass transit vehicle is in the PUDO state comprises determining a door position of a door of the mass transit vehicle, wherein the door position comprises at least one of an open position or a closed position.
9 . The system of claim 1 , wherein determining the mass transit vehicle is in the PUDO state comprises:
obtaining sensor data from a sensing system of the AV, wherein the sensor data comprises data characterizing a traffic sign of the mass transit vehicle; and
applying a MLM onboard the AV to the sensor data to generate an output indicating that the traffic sign indicates the mass transit vehicle is in the PUDO state.
10 . A system, comprising:
a memory; and
one or more processing devices, coupled to the memory, configured to perform operations comprising:
determining, based on map data obtained from a mapping system of an autonomous vehicle (AV), a distance of a cable car from a passenger pick-up/drop-off (PUDO) zone, wherein the cable car is located in an environment of the AV, and
determining, based at least on the distance, that the cable car is in a temporal vicinity of a PUDO state and that there is a lack of pedestrian activity involving boarding or disembarking from the cable car for a first threshold time interval, and modifying operation of the AV to at least partially circumnavigate the cable car.
11 . The system of claim 10 , wherein:
the operations further comprise determining an acceleration of the cable car; and
determining the cable car is in the temporal vicinity of the PUDO state is further based on the cable car being under a threshold acceleration.
12 . The system of claim 10 , wherein determining the cable car is in the temporal vicinity of the PUDO state further comprises:
obtaining, from sensor data, data characterizing a gaze of a pedestrian of the one or more pedestrians, wherein the pedestrian is located on an exterior portion of the cable car; and
generating, using a machine learning (ML) model onboard the AV and based on the data characterizing the gaze of the pedestrian, an output indicating the pedestrian is disembarking from the cable car.
13 . The system of claim 10 , wherein autonomously modifying the operation of the AV comprises stopping the AV at least a predetermined distance from the cable car.
14 . The system of claim 10 , wherein determining that there is a lack of pedestrian activity involving boarding or disembarking from the cable car further comprises determining a presence of one or more pedestrians in the environment of the AV and that no pedestrians of the one or more pedestrians have boarded or disembarked from the cable car for the first threshold time interval.
15 . The system of claim 10 , wherein the operations further comprise, responsive to determining that the cable car has been in the temporal vicinity of the PUDO state for more than a second threshold time interval, further autonomously modifying the operation of the AV to at least partially circumnavigate the cable car.
16 . The system of claim 10 , wherein:
sensor data comprises data characterizing a stop sign in the environment of the AV; and
determining the cable car is in the temporal vicinity of the PUDO state comprises determining that the stop sign is located at a side of the cable car.
17 . A method comprising:
identifying a mass transit vehicle in an environment of an autonomous vehicle (AV);
responsive to determining the mass transit vehicle is in a pick-up/drop-off (PUDO) state and that there is a lack of pedestrian activity involving boarding or disembarking from the mass transit vehicle for a threshold time interval, providing movement instructions to a planning system of the AV to at least partially circumnavigate the mass transit vehicle; and
modifying operation of the AV based on the movement instructions.
18 . The method of claim 17 , wherein identifying the mass transit vehicle comprises:
obtaining, based on sensor data from a sensing system of the AV, a distance between a pair of rails on which the mass transit vehicle is positioned; and
identifying the mass transit vehicle as a cable car based on the distance between the pair of rails and a reference distance.
19 . The method of claim 17 , wherein determining the mass transit vehicle is in the PUDO state comprises determining, based on map data, at least one of:
the mass transit vehicle is within a first threshold distance from a passenger PUDO zone; or
the mass transit vehicle is entering an intersection, and the passenger PUDO zone is within a second threshold distance from the intersection.
20 . The method of claim 17 , wherein determining the mass transit vehicle is in the PUDO state comprises:
obtaining sensor data from a sensing system of the AV, wherein the sensor data comprises data characterizing a gaze of a passenger of the mass transit vehicle; and
applying a machine learning model (MLM) onboard the AV to the sensor data to generate an output indicating that the gaze of the passenger indicates that the mass transit vehicle is in the PUDO state.