Method for transferring control of an autonomous vehicle to a remote operator

Abstract

One variation of a method for transferring control of an autonomous vehicle to a remote operator includes: accessing a specification for triggering manual control of autonomous vehicles; identifying a road segment, within a geographic region, exhibiting characteristics defined by the specification; and associating a location of the road segment, represented in a navigation map, with a remote operator trigger. The method also includes, at the autonomous vehicle operating within the geographic region: autonomously navigating along a route; transmitting a request for manual assistance to the remote operator in response to approaching the location associated with the remote operator trigger; transmitting sensor data to a remote operator portal associated with the remote operator; and executing a navigational command received from the remote operator via the remote operator portal; and resuming autonomous navigation along the route after passing the location.

I claim: 
     
         1 . A method for transferring control of an autonomous vehicle to a remote operator comprising:
 accessing a corpus of driving records of a fleet of autonomous vehicles operating within a geographic region;   identifying a road segment, within the geographic region, associated with a frequency of transitions, from autonomous operation to local manual operation triggered by local operators occupying autonomous vehicles in the fleet, that exceeds a threshold frequency based on the corpus of driving records;   associating a location of the road segment, represented in a navigation map, with a remote operator trigger; and   at the autonomous vehicle operating within the geographic region:
 autonomously navigating along a route; 
 in response to approaching the location associated with the remote operator trigger:
 transmitting a request for manual assistance to the remote operator; 
 transmitting sensor data to a remote operator portal associated with the remote operator; and 
 executing a navigational command received from the remote operator via the remote operator portal; and 
 
 resuming autonomous navigation along the route after passing the location. 
   
     
     
         2 . The method of  claim 1 :
 wherein accessing the corpus of driving records of the fleet of autonomous vehicles comprises accessing geospatial locations, within the geographic region, of instances of transition from autonomous operation to local manual operation triggered by local operators occupying autonomous vehicles in the fleet; and   wherein identifying the road segment comprises:
 aggregating instances of transition from autonomous operation to manual operation into a set of groups based on geospatial proximity, the set of groups comprising a first group containing instances of transition at geospatial locations along the road segment; 
 for the first group, calculating the frequency of transitions based on a ratio of quantity of transitions in the first group to quantity of instances of autonomous vehicles in the fleet traversing the road segment; and 
 flagging the road segment for the remote operator trigger in response to the frequency of transitions exceeding the threshold frequency. 
   
     
     
         3 . The method of  claim 2 :
 wherein accessing the corpus of driving records of the fleet of autonomous vehicles comprises accessing times of instances of transition from autonomous operation to manual operation;   wherein aggregating instances of transition from autonomous operation to manual operation into the set of groups comprises aggregating instances of transition from autonomous operation to manual operation into the set of groups further based on temporal proximity, the first group containing instances of transition at geospatial locations along the road segment within a daily time window;   further comprising limiting the remote operator trigger for the road segment according to the daily time window; and   wherein transmitting the request for manual assistance to the remote operator comprises, at the autonomous vehicle, transmitting the request for manual assistance to the remote operator in response to approaching the location of the remote operator trigger during the daily time window.   
     
     
         4 . The method of  claim 1 :
 wherein accessing the corpus of driving records of the fleet of autonomous vehicles comprises accessing geospatial locations, within the geographic region, of instances of transition from autonomous operation to local manual operation triggered by local operators occupying autonomous vehicles in the fleet and scene characteristics proximal autonomous vehicles during instances of transition from autonomous operation to manual operation;   wherein identifying the road segment comprises:
 aggregating instances of transition from autonomous operation to manual operation into a set of groups based on geospatial proximity and similarity of scene characteristics, the set of groups comprising a first group containing instances of transition at geospatial locations along the road segment and associated with a particular scene characteristic; and 
 flagging the road segment for the remote operator trigger in response to the frequency of transitions in the first group exceeding the threshold frequency; 
   further comprising limiting the remote operator trigger for the road segment according to the particular scene characteristic; and   wherein transmitting the request for manual assistance to the remote operator comprises, at the autonomous vehicle, transmitting the request for manual assistance to the remote operator in response to detecting the particular scene characteristic while approaching the location of the remote operator trigger.   
     
     
         5 . The method of  claim 4 :
 wherein accessing scene characteristics proximal autonomous vehicles during instances of transition from autonomous operation to manual operation comprises accessing scene characteristics comprising offsets between anteroposterior axes of autonomous vehicles and the Sun during instances of transition from autonomous operation to manual operation;   wherein identifying the road segment comprises identifying the first group containing instances of transition at geospatial locations along the road segment concurrent with offsets, between anteroposterior axes of autonomous vehicles and the Sun, within a solar offset window;   wherein limiting the remote operator trigger for the road segment according to the particular scene characteristic comprises limiting the remote operator trigger for the road segment according to the solar offset window; and   wherein transmitting the request for manual assistance to the remote operator comprises, at the autonomous vehicle, transmitting the request for manual assistance to the remote operator in response to an offset between an anteroposterior axes of the autonomous vehicle and the Sun falling within the solar offset window while approaching the location of the remote operator trigger.   
     
     
         6 . The method of  claim 4 :
 wherein accessing scene characteristics proximal autonomous vehicles during instances of transition from autonomous operation to manual operation comprises accessing scene characteristics comprising presence of pedestrians proximal autonomous vehicles during instances of transition from autonomous operation to manual operation;   wherein identifying the road segment comprises identifying the first group containing instances of transition at geospatial locations along the road segment concurrent with presence of a minimum quantity of pedestrians;   wherein limiting the remote operator trigger for the road segment according to the particular scene characteristic comprises associating the remote operator trigger for the road segment with presence of the minimum quantity of pedestrians; and   wherein transmitting the request for manual assistance to the remote operator comprises, at the autonomous vehicle, transmitting the request for manual assistance to the remote operator in response to detecting more than the minimum quantity of pedestrians proximal the autonomous vehicle while approaching the location of the remote operator trigger.   
     
     
         7 . The method of  claim 1 :
 wherein autonomously navigating along the route comprises, at the autonomous vehicle, autonomously navigating from a pickup location to a drop-off location specified by a user while the user occupies the autonomous vehicle and while a local operator is absent from the autonomous vehicle;   wherein transmitting the request for manual assistance to the remote operator comprises transmitting the request for manual assistance to the remote operator in response to approaching the location associated with the remote operator trigger while a local operator is absent from the autonomous vehicle; and   further comprising, at a second autonomous vehicle occupied by a second local operator:
 autonomously navigating along a second route; 
 in response to approaching the location associated with the remote operator trigger while occupied by the second local operator, prompting the second local operator to assume manual control of the second autonomous vehicle; and 
 in response to passing the location, prompting the local operator to confirm autonomous navigation of the second autonomous vehicle along the second route. 
   
     
     
         8 . The method of  claim 1 :
 wherein transmitting the request for manual assistance to the remote operator comprises, transmitting the request for manual assistance to a remote operator manager in response to approaching the location associated with the remote operator trigger;   at the remote operator manager:
 in response to receiving the request for manual assistance from the autonomous vehicle, selecting the remote operator from a set of available remote operators; and 
 routing sensor data received from the autonomous vehicle to the remote operator portal, associated with the remote operator, via a computer network. 
   
     
     
         9 . The method of  claim 1 :
 further comprising, at the autonomous vehicle, estimating a time of arrival of the autonomous vehicle at the location associated with the remote operator trigger while autonomously navigating along the route;   wherein transmitting the request for manual assistance to the remote operator comprises, at the autonomous vehicle, transmitting the request for manual assistance to the remote operator in response to the time of arrival falling below a threshold duration at a first time;   wherein transmitting sensor data to the remote operator portal comprises, at the autonomous vehicle, streaming sensor data to the remote operator portal in real-time over a wireless computer network following the first time; and   further comprising ceasing transmission of sensor data to the remote operator portal after passing the location.   
     
     
         10 . The method of  claim 1 :
 further comprising autonomously slowing to a stop in response to approaching the location associated with the remote operator trigger;   wherein transmitting the request for manual assistance to the remote operator comprises transmitting the request for manual confirmation to resume autonomous navigation along the route; and   wherein executing the navigational command received from the remote operator comprises resuming autonomous navigation along the route past the location in response to receipt of manual confirmation from the remote operator.   
     
     
         11 . method of  claim 1 :
 further comprising, in response to confirmation of manual assistance from the remote operator, transferring braking, acceleration, and steering controls of the autonomous vehicle to the remote operator portal;   wherein executing the navigational command received from the remote operator comprises executing braking, acceleration, and steering commands received from the remote operator portal; and   wherein resuming autonomous navigation along the route comprises resuming autonomous navigation along the route in response to receipt of confirmation from the remote operator to resume autonomous navigation.   
     
     
         12 . The method of  claim 1 , further comprising:
 accessing a corpus of historical traffic accident data of manually-operated vehicles involved in traffic accidents within the geographic region;   identifying a second road segment, within the geographic region, associated with a frequency of traffic accidents that exceeds a second threshold frequency based on the corpus of historical traffic accident data;   associating a second location of the second road segment, in the navigation map, with a second remote operator trigger; and   at the autonomous vehicle:
 autonomously navigating along a second route; 
 in response to approaching the second location associated with the second remote operator trigger:
 transmitting a second request for manual assistance to a second remote operator; 
 transmitting sensor data to a second remote operator portal associated with the second remote operator; and 
 executing a second navigational command received from the second remote operator via the second remote operator portal; and 
 
 resuming autonomous navigation along the second route after passing the second location. 
   
     
     
         13 . The method of  claim 1 :
 wherein autonomously navigating along the route comprises, at the autonomous vehicle, autonomously navigating along the route based on an autonomous navigation model;   further comprising, at a remote computer system:
 recording a corpus of sensor data received from the autonomous vehicle following the request for manual assistance; 
 recording the navigational command entered by the remote operator and served to the autonomous vehicle responsive to the request for manual assistance; 
 generating a revised autonomous navigation model based on the corpus of sensor data and the navigational command; and 
 loading the revised autonomous navigation model onto the autonomous vehicle. 
   
     
     
         14 . The method of  claim 1 , further comprising:
 deriving a set of characteristics of the road segment;   scanning the navigation map for a second road segment exhibiting characteristics similar to the set of characteristics of the road segment;   associating a second location of the second road segment, in the navigation map, with a second remote operator trigger; and   at the autonomous vehicle:
 autonomously navigating along a second route; 
 in response to approaching the second location associated with the second remote operator trigger:
 transmitting a second request for manual assistance to a second remote operator; 
 transmitting sensor data to a second remote operator portal associated with the second remote operator; and 
 executing a second navigational command received from the second remote operator via the second remote operator portal; and 
 
 resuming autonomous navigation along the second route after passing the second location. 
   
     
     
         15 . The method of  claim 1 , further comprising:
 identifying a second road segment, within the geographic region, associated with a second frequency of transitions, from autonomous operation to remote manual operation triggered by autonomous vehicles in the fleet, that exceeds a second threshold frequency based on the corpus of driving records;   associating a second location of the second road segment, represented in the navigation map, with a second remote operator trigger; and   at the autonomous vehicle:
 autonomously navigating along a second route; 
 in response to approaching the second location associated with the second remote operator trigger:
 transmitting a second request for manual assistance to a second remote operator; 
 transmitting sensor data to a second remote operator portal associated with the second remote operator; and 
 executing a second navigational command received from the second remote operator via the second remote operator portal; and 
 
 resuming autonomous navigation along the second route after passing the second location. 
   
     
     
         16 . A method for transferring control of an autonomous vehicle to a remote operator comprising:
 accessing a corpus of historical traffic accident data of manually-operated vehicles involved in traffic accidents within a geographic region;   identifying a road segment, within the geographic region, associated with a frequency of traffic accidents that exceeds a threshold frequency based on the corpus of historical traffic accident data;   associating a location of the road segment, represented in a navigation map, with a remote operator trigger; and   at the autonomous vehicle operating within the geographic region:
 autonomously navigating along a route; 
 in response to approaching the location associated with the remote operator trigger:
 transmitting a request for manual assistance to the remote operator; 
 transmitting sensor data to a remote operator portal associated with the remote operator; and 
 executing a navigational command received from the remote operator via the remote operator portal; and 
 
 resuming autonomous navigation along the route after passing the location. 
   
     
     
         17 . A method for transferring control of an autonomous vehicle to a remote operator comprising:
 accessing a specification for triggering manual control of autonomous vehicles;   identifying a road segment, within a geographic region, exhibiting characteristics defined by the specification;   associating a location of the road segment, represented in a navigation map, with a remote operator trigger; and   at the autonomous vehicle operating within the geographic region:
 autonomously navigating along a route; 
 in response to approaching the location associated with the remote operator trigger:
 transmitting a request for manual assistance to the remote operator; 
 transmitting sensor data to a remote operator portal associated with the remote operator; and 
 executing a navigational command received from the remote operator via the remote operator portal; and 
 
 resuming autonomous navigation along the route after passing the location. 
   
     
     
         18 . The method of  claim 17 :
 wherein accessing the specification for triggering manual control of autonomous vehicles comprises accessing a threshold frequency of traffic accidents per unit distance;   wherein identifying the road segment comprises:
 accessing a corpus of historical traffic accident data of manually-operated vehicles involved in traffic accidents within the geographic region; 
 based on the corpus of historical traffic accident data, isolating the road segment associated with a frequency of historical traffic accidents exceeding the threshold frequency. 
   
     
     
         19 . The method of  claim 17 :
 wherein accessing the specification for triggering manual control of autonomous vehicles comprises accessing a threshold frequency of transitions, from autonomous operation to remote manual operation, triggered by autonomous vehicles;   wherein identifying the road segment comprises:
 accessing a corpus of driving records of a fleet of autonomous vehicles operating within the geographic region; 
 isolating the road segment, within the geographic region, associated with a frequency of transitions, from autonomous operation to local manual operation triggered by autonomous vehicles in the fleet, that exceeds the threshold frequency of transitions based on the corpus of driving records. 
   
     
     
         20 . The method of  claim 17 :
 wherein accessing the specification for triggering manual control of autonomous vehicles comprises accessing a threshold frequency of transitions, from autonomous operation to remote manual operation, triggered by local operators occupying autonomous vehicles;   wherein identifying the road segment comprises:
 accessing a corpus of driving records of a fleet of autonomous vehicles operating within the geographic region; 
 isolating the road segment, within the geographic region, associated with a frequency of transitions, from autonomous operation to local manual operation triggered by local operators occupying autonomous vehicles in the fleet, that exceeds the threshold frequency of transitions based on the corpus of driving records.