Communications for autonomous vehicles
Aspects of the disclosure provide a method of facilitating communications from an autonomous vehicle to a user. For instance, a method may include, while attempting to pick up the user and prior to the user entering an vehicle, inputting a current location of the vehicle and map information into a model in order to identify a type of communication action for communicating a location of the vehicle to the user; enabling a first communication based on the type of the communication action; determining whether the user has responded to the first communication from received sensor data; and enabling a second communication based on the determination of whether the user has responded to the communication.
1. A method of training a model for facilitating communications from an autonomous vehicle to a user, the method including:
receiving, by one or more computing devices, training data including a first training input that indicates a location of a vehicle, a second training input that indicates map information, a third training input that indicates a location of the user, a fourth training input that characterizes sensor data identifying one or more objects in the vehicle's environment, and a target output indicating a type of communication;
training the model, by the one or more computing devices, on the training data according to current values of parameters of the model to generate a set of output values indicative of one or more types of communications;
determining a difference value using the target output and the set of output values; and
adjusting the current values of parameters of the model using the difference value.
2. The method of claim 1 , wherein the training data corresponds to a request by the user to cause the vehicle to perform the type of communication in order to communicate with the user.
3. The method of claim 1 , wherein the type of communication is an audible communication.
4. The method of claim 1 , wherein the type of communication is a visual communication.
5. The method of claim 1 , wherein the training data further includes ambient lighting conditions.
6. The method of claim 1 , wherein the set of output values is further indicative of a level of appropriateness for a type of communication.
7. The method of claim 1 , wherein the determining includes comparing the target output to the set of output values.
8. The method of claim 1 , wherein the second training input is used to train the model to determine which one of a plurality of communication options should be provided or enabled, the plurality of communication options including at least one of flashing the vehicle's headlights, honking a horn of the vehicle, or displaying information on an electronic display of the vehicle.
9. The method of claim 1 , further comprising receiving, by one or more computing devices, training data including ambient lighting conditions in order to train the model to differentiate between different ambient lighting conditions for different types of communications output by the model.
10. The method of claim 1 , further comprising assigning weights for different communication options.
11. The method of claim 10 , wherein the weights are assigned based on at least one of the map information or the sensor data.
12. The method of claim 1 , wherein the model is trained to differentiate between situations in which a visual communication is appropriate versus when an audible communication is appropriate.
13. The method of claim 12 , wherein training the model is further based on user feedback indicating whether a particular communication was inappropriate or inconvenient.
14. The method of claim 13 , wherein the feedback further indicates a reason why the particular communication was inappropriate or inconvenient.
15. The method of claim 1 , wherein training the model is further based on a user's history of being picked up at a particular location.
16. The method of claim 1 , wherein the map information includes at least one of distance to a nearest curb, staircase, entrance or exit.
17. The method of claim 1 , wherein the map information indicates whether the vehicle is proximate to another object that may obstruct a user's view of the vehicle.
18. An autonomous vehicle, comprising:
a memory; and
one or more computing devices operably coupled with the memory, the one or more computing devices configured to:
receive training data including a first training input that indicates a location of a vehicle, a second training input that indicates map information, a third training input that indicates a location of a user, a fourth training input that characterizes sensor data identifying one or more objects in the vehicle's environment, and a target output indicating a type of communication;
train the model, by the one or more computing devices, on the training data according to current values of parameters of the model to generate a set of output values indicative of one or more types of communications;
determine a difference value using the target output and the set of output values; and
adjust the current values of parameters of the model using the difference value.
19. The autonomous vehicle of claim 18 , wherein the training data corresponds to a request by the user to cause the vehicle to perform the type of communication in order to communicate with the user, and
wherein the type of communication is at least one of an audible communication or a visual communication.
20. The autonomous vehicle of claim 18 , wherein the set of output values is further indicative of a level of appropriateness for a type of communication.