System and method for computing automatic notifications about a policy of behavior generated by an artificial intelligence algorithm

Patent metadata

Field	Value
Publication number	US-12344266-B2
Application number	US-202318316492-A
Country	US
Kind code	B2
Filing date	May 12, 2023
Priority date	May 12, 2023
Publication date	Jul 1, 2025
Grant date	Jul 1, 2025

Abstract

A vehicle capable of operating autonomously, a system for generating a notification for a behavior of the vehicle includes a method of generating the notification using artificial intelligence. The system includes an interface and a processor. The processor is configured to determine a current state of the vehicle, determine probabilities for a plurality of trajectories through a behavior model of the vehicle, each trajectory indicating one or more decisions from the current state to a subsequent state of the vehicle using the behavior model for the vehicle, generate a message to be presented as the vehicle performs the behavior that is determined by the artificial intelligence system, wherein the behavior is selected based on the probabilities, determine a temporal parameter for the message, determine a modality parameter for the message, and present the message at the interface using the temporal parameter and the modality parameter.

First claim

What is claimed is:

1. A method of generating a notification for a behavior of a vehicle computed using an artificial intelligence system to operate the vehicle autonomously or semi-autonomously, comprising:
determining a current state of the vehicle;
creating a behavior model of the vehicle by a model-based planning artificial intelligence solution that includes a decision tree and a plurality of trajectories, each trajectory including a plurality of nodes of the decision tree between the current state and a subsequent state;
determining probabilities for the plurality of trajectories through the behavior model, wherein the probability of each trajectory is based on the plurality of nodes between the current state and the subsequent state;
selecting a sub-trajectory of the trajectory, the sub-trajectory including a subset of the plurality of nodes between the current state and the subsequent state;
selecting a significant action of the sub-trajectory;
generating a message based on the significant action, the message to be presented as the vehicle performs a behavior that is determined by the artificial intelligence system, wherein the behavior is selected based on the probabilities;
determining a temporal parameter for the message;
determining a modality parameter for the message; and
presenting the message using the temporal parameter and the modality parameter.

2. The method of claim 1 , wherein the behavior model is created using a model-based planning artificial intelligence solution that is evaluated offline to produce the probabilities of transitions between states.

3. The method of claim 1 , further comprising selecting a set of optimal trajectories from the plurality of trajectories and a set of optimal behaviors associated with the set of optimal trajectories based on a probability of the set of optimal trajectories.

4. The method of claim 1 , wherein the temporal parameter is selected using one of: (i) a pre-determined rule; (ii) a confidence value determined using artificial intelligence; (iii) a value of an action; and (iv) an entropy value.

5. The method of claim 1 , wherein the message explains a reason for the behavior of the vehicle, wherein the reason is based on at least one of: (i) an action taken by the vehicle; (ii) a sequence of actions taken by the vehicle; (iii) a feature of the action taken by the vehicle; (iv) a feature of a state of the vehicle when the action is taken by the vehicle;
(v) a value of the action taken by the vehicle; and (vi) the action not taken by the vehicle.

6. A system for generating a notification for a behavior of a vehicle computed using an artificial intelligence system to operate the vehicle autonomously, comprising:
an interface;
a processor configured to:
determine a current state of the vehicle;
create a behavior model of the vehicle by a model-based planning artificial intelligence solution that includes a decision tree and a plurality of trajectories, each trajectory including a plurality of nodes of the decision tree between the current state and a subsequent state;
determine probabilities for the plurality of trajectories through behavior model, wherein the probability of each trajectory is based on the plurality of nodes between the current state and the subsequent state;
select a sub-trajectory of the trajectory, the sub-trajectory including a subset of the plurality of nodes between the current state and the subsequent state;
select a significant action of the sub-trajectory;
generate a message based on the significant action, the message to be presented as the vehicle performs the behavior that is determined by the artificial intelligence system, wherein the behavior is selected based on the probabilities;
determine a temporal parameter for the message;
determine a modality parameter for the message; and
present the message at the interface using the temporal parameter and the modality parameter.

7. The system of claim 6 , wherein the processor is further configured to create the behavior model using a model-based planning artificial intelligence solution that is evaluated offline to produce the probabilities of transitions between states.

8. The system of claim 6 , wherein the processor is further configured to select a set of optimal trajectories from the plurality of trajectories and a set of optimal behaviors associated with the set of optimal trajectories based on a probability of the set of optimal trajectories.

9. The system of claim 8 , wherein the processor is further configured to select the temporal parameter using one of: (i) a pre-determined rule; (ii) a confidence value determined using artificial intelligence; (iii) a value of an action; and (iv) an entropy value.

10. The system of claim 6 , wherein the message explains a reason for the behavior of the vehicle, wherein the reason is based on at least one of: (i) an action taken by the vehicle; (ii) a sequence of actions taken by the vehicle; (iii) a feature of the action taken by the vehicle; (iv) a feature of a state of the vehicle when the action is taken by the vehicle;
(v) a value of the action taken by the vehicle; and (vi) the action not taken by the vehicle.

11. A vehicle capable of operating autonomously, comprising:
an interface;
a processor configured to:
determine a current state of the vehicle;
create a behavior model of the vehicle by a model-based planning artificial intelligence solution that includes a decision tree and a plurality of trajectories, each trajectory including a plurality of nodes of the decision tree between the current state and a subsequent state;
determine probabilities for the plurality of trajectories through the behavior model, wherein the probability of each trajectory is based on the plurality of nodes between the current state and the subsequent state;
select a sub-trajectory of the trajectory, the sub-trajectory including a subset of the plurality of nodes between the current state and the subsequent state;
select a significant action of the sub-trajectory;
generate a message based on the significant action, the message to be presented as the vehicle performs a behavior that is determined by an artificial intelligence system, wherein the behavior is selected based on the probabilities;
determine a temporal parameter for the message;
determine a modality parameter for the message; and
present the message at the interface using the temporal parameter and the modality parameter.

12. The vehicle of claim 11 , wherein the processor is further configured to create the behavior model using a model-based planning artificial intelligence solution that is evaluated offline to produce the probabilities of transitions between states.

13. The vehicle of claim 11 , wherein the processor is further configured to select a set of optimal trajectories from the plurality of trajectories and a set of optimal behaviors associated with the set of optimal trajectories based on a probability of the set of optimal trajectories.

14. The vehicle of claim 13 , wherein the message explains a reason for the behavior of the vehicle, wherein the reason is based on at least one of: (i) an action taken by the vehicle; (ii) a sequence of actions taken by the vehicle; (iii) a feature of the action taken by the vehicle; (iv) a feature of a state of the vehicle when the action is taken by the vehicle;
(v) a value of the action taken by the vehicle; and (vi) the action not taken by the vehicle.

15. The method of claim 1 , wherein each trajectory indicates one or more decisions from the current state to the subsequent state using the behavior mo

Assignees

• Gm Global Tech Operations Llc

Inventors

• Goldman-Shenhar Claudia
• Bustin Ronit

Classifications

• G06N5/01 Primary

  Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound · CPC title

• B60W2050/143

  Alarm means · CPC title

• B60W2050/146

  Display means · CPC title

• B60W60/0011

  involving control alternatives for a single driving scenario, e.g. planning several paths to avoid obstacles · CPC title

• B60W50/14 Primary

  Means for informing the driver, warning the driver or prompting a driver intervention · CPC title