System and method for controlling a vehicle using contextual navigation assistance

1. A system for controlling a vehicle traversing a vehicle transportation network using contextual navigation assistance, the system comprising:

a processor configured to receive, while the vehicle traverses a route from a starting location to a stopping location within the vehicle transportation network, sensor data indicating at least a condition of the vehicle;

a context model configured to define a context for the vehicle using the sensor data for a plurality of time points while the vehicle traverses from the starting location to the stopping location, wherein the context is one of multiple contexts, and each context is associated with at least one candidate variable for inclusion in an ordered representation of information;

a phase classifier configured to classify, using the sensor data, the vehicle into a phase of multiple phases associated with different proximities of the vehicle to a maneuver along the route for the plurality of time points, wherein each phase is associated with at least one candidate variable for inclusion in the ordered representation of information;

an information-needs model configured to generate, using the phase and the context as input, the ordered representation of information regarding the maneuver of the vehicle along the route for the plurality of time points, wherein the ordered representation of information is ranked from most likely information that is necessary, during a current phase of the multiple phases, to successfully complete the maneuver to least likely information that is necessary, during the current phase of the multiple phases, to successfully complete the maneuver;

a user interface configured to produce the ordered representation of information at a current time point for use during the current phase of the multiple phases of the maneuver; and

a trajectory controller configured to control at least one component of the vehicle during the current phase of the multiple phases based on feedback from the vehicle related to at least one entry in the ordered representation of information.

2. The system of claim 1 , wherein:

the sensor data comprises sensor data from at least one sensor monitoring an environment external to the vehicle and from at least one sensor monitoring an environment internal to the vehicle;

the sensor data for the environment external to the vehicle includes at least two of Global Positioning System data, camera data from an externally-facing camera, radar data, lidar data, or network data made available to the vehicle from other vehicles within the vehicle transportation network; and

the sensor data for the environment internal to the vehicle includes at least two of camera data from an internally-facing camera monitoring a driver of the vehicle, sound data from a microphone within the vehicle, steering data indicating motion of a steered wheel of the vehicle, or pedal data indicating motion of one or more pedals of the vehicle.

3. The system of claim 2 , wherein:

an operator of the vehicle is the driver of the vehicle;

the user interface comprises at least one of a speaker or a video monitor within the vehicle; and

the feedback comprises a change in the sensor data.

4. The system of claim 1 , wherein the phase classifier is configured to classify the vehicle into the phase of multiple phases by:

determining a time-distance from the maneuver, wherein the maneuver is a current maneuver;

determining a position of the vehicle relative to a previous maneuver before the current maneuver;

determining whether or not the current maneuver is within a line-of-sight of the vehicle; and

classifying the vehicle into the phase using the time-distance from the current maneuver, the position of the vehicle relative to the previous maneuver, and whether or not the current maneuver is within a line-of-sight of the vehicle as inputs into a decision tree whose output is the phase.

5. The system of claim 1 , wherein:

at least one of the multiple phases is divided into a plurality of subphases, and

the phase classifier is configured to classify the vehicle into the phase of multiple phases by:

classifying the vehicle into the phase of the multiple phases; and

upon a condition that the phase is divided into a plurality of subphases, classifying the vehicle into a subphase of the phase.

6. The system of claim 1 , wherein the context model comprises:

a first neural network using map data, external camera data, and location data as inputs, and outputting external context features;

a second neural network using at least two of internal camera data, wheel data, steering data, input data from an operator of the vehicle, or previous outputs to the operator as inputs, and outputting internal context features; and

a decision tree that receives the external context features and the internal context feature as input and outputs the context.

7. The system of method of claim 6 , wherein:

the second neural network uses at least two of the internal camera data, the wheel data, the steering data, the input data from an operator of the vehicle, the previous outputs to the operator, or network data as inputs, the network data including at least one of live traffic data or operator status data of a driver within the vehicle transportation network.

8. A system for controlling a vehicle traversing a vehicle transportation network using contextual navigation assistance, the system comprising:

a processor configured to:

receive, while the vehicle traverses a route from a starting location to a stopping location within the vehicle transportation network, sensor data indicating at least a condition of the vehicle;

for a plurality of time points while the vehicle traverses from the starting location to the stopping location:

generate, responsive to the sensor data, an ordered representation of information regarding a maneuver of the vehicle along the route, wherein the maneuver comprises multiple phases associated with different proximities of the vehicle to the maneuver, and the ordered representation of information is ranked from most likely information that is necessary, during a current phase of the multiple phases, to successfully complete the maneuver to least likely information that is necessary, during the current phase of the multiple phases, to successfully complete the maneuver; and

produce, by a user interface, the ordered representation of information at a current time point for use in controlling the vehicle during the current phase of the multiple phases of the maneuver.

9. The system of claim 8 , further comprising:

a trajectory controller configured to control at least one component of the vehicle during a current phase of the multiple phases based on feedback from the vehicle related to at least one entry in the ordered representation of information.

10. The system of claim 8 , wherein the processor is configured to, for a plurality of time points while the vehicle traverses from the starting location to the stopping location:

determine the current phase of the multiple phases using a phase classifier, the phase classifier comprising:

a preprocessing component that uses the route and a Global Positioning System location of the vehicle to determine intermediate data comprising at least a time-distance of the vehicle from the maneuver; and

a decision tree receiving the intermediate data and outputting the current phase.

11. The system of claim 10 , wherein the maneuver is a current maneuver, and the intermediate data comprises:

a position of the vehicle relative to a maneuver before the current maneuver.

12. The system of claim 8 , wherein the processor is configured to:

transmit, based on the context, network data for use by other vehicles within the vehicle transportation network, the network data including at least one of live traffic data or operator status data of a driver within the vehicle.

13. A method for controlling a vehicle traversing a vehicle transportation network using contextual navigation assistance, the method comprising:

receiving, while the vehicle traverses a route from a starting location to a stopping location within the vehicle transportation network, sensor data indicating at least a condition of the vehicle;

generating, responsive to the sensor data, a first ordered representation of information regarding a maneuver of the vehicle along the route, wherein the maneuver comprises multiple phases associated with different proximities of the vehicle to the maneuver, and the first ordered representation of information is ranked from most likely information necessary, during a first phase of the multiple phases, to successfully complete the maneuver to least likely information that is necessary, during the first phase of the multiple phases, to successfully complete the maneuver;

producing, by a user interface, the first ordered representation of information at a first point in time for use while during the first phase of the multiple phases of the maneuver;

controlling, using a trajectory controller, the vehicle during the first phase based on feedback from the vehicle related to at least one entry in the first ordered representation of information;

generating, responsive to the sensor data, a second ordered representation of information regarding the maneuver of the vehicle along the route, wherein the second ordered representation of information is ranked from most likely information that is necessary to know, during a second phase of the multiple phases, to successfully complete the maneuver to least likely information that is necessary to know, during the second phase of the multiple phases, to successfully complete the maneuver, and the second phase of the multiple phases is subsequent to the first phase;

producing, by the user interface, the second ordered representation of information at a second point in time for use during the second phase of the multiple phases, wherein the second point in time is subsequent to the first point in time; and

controlling, using a trajectory controller, the vehicle during the second phase based on feedback from the vehicle related to at least one entry in the second ordered representation of information.

14. The method of claim 13 , wherein generating the second ordered representation is responsive to the sensor data and the feedback from the vehicle during the first phase of the multiple phases.

15. The method of claim 13 , further comprising:

classifying, by a phase classifier, the vehicle into a phase of the multiple phases using the sensor data; and

providing the phase to an information-needs model, the information-needs model generating multiple ordered representations including the first ordered representation and the second ordered representation.

16. The method of claim 15 , wherein classifying the vehicle into the phase of the multiple phases comprises:

determining a time-distance from the maneuver, wherein the maneuver is a current maneuver;

determining a position of the vehicle relative to a previous maneuver before the current maneuver; and

classifying the vehicle into the phase using the time-distance from the current maneuver and the position of the vehicle relative to the previous maneuver.

17. The method of claim 13 , further comprising:

classifying, by a phase classifier, the vehicle into each of the first phase and the second phase using a current position of the vehicle, the route of the vehicle, and a relationship between the current position and at least two adjacent maneuvers in the route;

providing the first phase as input for generating the first ordered representation of information; and

providing the second phase as input for generating the second ordered representation of information.

18. The method of claim 13 , wherein:

the first phase is divided into a plurality of first subphases,

the first ordered representation of information is ranked from the most likely information that is necessary, during the first phase and a first subphase of the plurality of first subphases, to successfully complete the maneuver to the least likely information that is necessary, during the first phase and the first subphase, to successfully complete the maneuver;

the second phase is divided into a plurality of second subphases, and

the second ordered representation of information is ranked from the most likely information that is necessary, during the second phase and a second subphase of the plurality of second subphases, to successfully complete the maneuver to the least likely information that is necessary, during the second phase and the second subphase, to successfully complete the maneuver.

19. The method of claim 13 , further comprising:

defining, by a context model, a context for the vehicle using the sensor data, wherein:

the context is one of multiple contexts including a first context and a second context,

each context is associated with at least one candidate variable for inclusion in an ordered representation of information,

the first context is associated with at least one candidate variable for inclusion in the first ordered representation of information, and

the second context is associated with at least one candidate variable for inclusion in the second ordered representation of information; and

providing the context to an information-needs model, the information-needs model generating multiple ordered representations including the first ordered representation and the second ordered representation.

20. The method of claim 19 , wherein the context model comprises:

at least one classifier that receives the sensor data as input and identifies individual features using the sensor data, each feature related to at least one of an internal condition or an external condition of the vehicle; and

a decision tree that receives the individual features as input and outputs the context.