Granted Patent B2

US 12,444,244 · App. 18/145,557 · Granted Oct 14, 2025

Inventors: Dong Li (Beijing, CN); Bin Wang (Beijing, CN); Wulong Liu (Montreal, CA); Yuzheng Zhuang (Shenzhen, CN)

Assignee: Huawei Technologies Co., Ltd.

G07C5/02B60W50/06G06N7/01B60W2050/0018B60W60/001

Abstract

Claims (73)

1. A driving decision-making method, comprising:

obtaining, by an autonomous driving vehicle, information of a current driving environment state of the autonomous driving vehicle, wherein the autonomous driving vehicle includes one or more sensors;

constructing, by the autonomous driving vehicle, a Monte Carlo tree based on the current driving environment state, wherein the Monte Carlo tree comprises N nodes, each node represents a corresponding driving environment state, the N nodes comprise a root node and N−1 non-root nodes, the root node represents the current driving environment state, a first driving environment state represented by a first node is predicted by using a stochastic model of driving environments based on a second driving environment state represented by a parent node of the first node and based on a driving action, the driving action is determined by the parent node of the first node in a process of obtaining the first node through expansion, the first node is any node of the N−1 non-root nodes, and N is a positive integer greater than or equal to 2;

determining, by the autonomous driving vehicle, in the Monte Carlo tree based on at least one of an access count or a value function of each node in the Monte Carlo tree, a node sequence, wherein the node sequence comprises a plurality of nodes that starts from the root node and ends at a leaf node;

in response to determining the node sequence, determining, by the autonomous driving vehicle, a driving action sequence of a plurality of future driving steps, wherein each future driving step in the driving action sequence comprises a driving action corresponding to each node comprised in the node sequence, and wherein the driving action sequence is used by the autonomous driving vehicle for driving decision-making;

autonomously driving, by the autonomous driving vehicle, the autonomous driving vehicle based on a first driving action in the driving action sequence;

obtaining, by the autonomous driving vehicle, an actual driving environment state after the first driving action is executed; and

updating, by the autonomous driving vehicle, the stochastic model of driving environments based on the current driving environment state, the first driving action, and the actual driving environment state, wherein

the access count of each node is determined based on access counts of subnodes of the each node and an initial access count of the each node, the value function of the each node is determined based on value functions of subnodes of the each node and an initial value function of the each node, the initial access count of the each node is 1, and the initial value function of the each node is determined based on a value function that matches the corresponding driving environment state represented by the each node.

2. The method according to claim 1 , wherein that the first driving environment state represented by the first node is predicted by using the stochastic model of driving environments based on the second driving environment state represented by the parent node of the first node and based on the driving action comprises:

predicting, through dropout-based forward propagation by using the stochastic model of driving environments, a probability distribution of a driving environment state after the driving action is executed based on the second driving environment state represented by the parent node of the first node; and

obtaining the first driving environment state represented by the first node through sampling from the probability distribution.

3. The method according to claim 1 , wherein that the initial value function of the node is determined based on the value function that matches the driving environment state represented by the node comprises:

selecting, from an episodic memory, a first quantity of target driving environment states that have a highest matching degree with the driving environment state represented by the node; and

determining the initial value function of the node based on value functions respectively corresponding to the first quantity of target driving environment states.

4. The method according to claim 3 , wherein the method further comprises:

when a driving episode ends, determining a cumulative reward return value corresponding to an actual driving environment state after each driving action in the driving episode is executed; and

updating the episodic memory by using, as a value function corresponding to the actual driving environment state, the cumulative reward return value corresponding to the actual driving environment state after each driving action is executed.

5. The method according to claim 1 , wherein the node sequence is determined:

based on the access count of the each node in the Monte Carlo tree according to a maximum access count rule;

based on the value function of the each node in the Monte Carlo tree according to a maximum value function rule; or

based on the access count and the value function of the each node in the Monte Carlo tree according to a “maximum access count first, maximum value function next” rule.

6. The method according to claim 1 , wherein obtaining, by the autonomous driving vehicle, information of the current driving environment state of the autonomous driving vehicle comprises:

receiving, from a vehicle velocity sensor of the autonomous driving vehicle, a velocity of the autonomous driving vehicle.

7. The method according to claim 1 , obtaining, by the autonomous driving vehicle, information of the current driving environment state of the autonomous driving vehicle comprises:

receiving, from an acceleration sensor of the autonomous driving vehicle, an acceleration of the autonomous driving vehicle.

8. The method according to claim 1 , wherein obtaining, by the autonomous driving vehicle, information of the current driving environment state of the autonomous driving vehicle comprises:

receiving, from a distance sensor of the autonomous driving vehicle, a relative distance between the autonomous driving vehicle and another vehicle.

9. A driving decision-making apparatus in an autonomous driving vehicle, comprising:

at least one processor; and

one or more memories coupled to the at least one processor and storing programming instructions for execution by the at least one processor to:

obtain information of a current driving environment state of the autonomous driving vehicle, wherein the autonomous driving vehicle includes one or more sensors;

construct a Monte Carlo tree based on the current driving environment state, wherein the Monte Carlo tree comprises N nodes, each node represents a corresponding driving environment state, the N nodes comprise a root node and N−1 non-root nodes, the root node represents the current driving environment state, a first driving environment state represented by a first node is predicted by using a stochastic model of driving environments based on a second driving environment state represented by a parent node of the first node and based on a driving action, the driving action is determined by the parent node of the first node in a process of obtaining the first node through expansion, the first node is any node of the N−1 non-root nodes, and N is a positive integer greater than or equal to 2;

determine, in the Monte Carlo tree based on at least one of an access count or a value function of each node in the Monte Carlo tree, a node sequence, wherein the node sequence comprises a plurality of nodes that starts from the root node and ends at a leaf node;

in response to determining the node sequence, determine a driving action sequence of a plurality of future driving steps, wherein each driving future step in the driving action sequence comprises a driving action corresponding to each node comprised in the node sequence, and wherein the driving action sequence is used for driving decision-making;

autonomously drive the autonomous driving vehicle based on a first driving action in the driving action sequence;

obtain an actual driving environment state after the first driving action is executed; and

update the stochastic model of driving environments based on the current driving environment state, the first driving action, and the actual driving environment state, wherein

the access count of each node is determined based on access counts of subnodes of the each node and an initial access count of the each node, the value function of each node is determined based on value functions of subnodes of the each node and an initial value function of the each node, the initial access count of the each node is 1, and the initial value function of the each node is determined based on a value function that matches the corresponding driving environment state represented by the each node.

10. The apparatus according to claim 9 , wherein the programming instructions are for execution by the at least one processor to:

predict, through dropout-based forward propagation by using the stochastic model of driving environments, a probability distribution of a driving environment state after the driving action is executed based on the second driving environment state represented by the parent node of the first node; and

obtain the first driving environment state represented by the first node through sampling from the probability distribution.

11. The apparatus according to claim 9 , wherein the programming instructions are for execution by the at least one processor to:

select, from an episodic memory, a first quantity of target driving environment states that have a highest matching degree with the driving environment state represented by the node; and determine the initial value function of the node based on value functions respectively corresponding to the first quantity of target driving environment states.

12. The apparatus according to claim 11 , wherein the programming instructions are for execution by the at least one processor to:

when a driving episode ends, determine a cumulative reward return value corresponding to an actual driving environment state after each driving action in the driving episode is executed; and

update the episodic memory by using, as a value function corresponding to the actual driving environment state, the cumulative reward return value corresponding to the actual driving environment state after each driving action is executed.

13. The apparatus according to claim 9 , wherein the node sequence is determined:

based on the access count of the each node in the Monte Carlo tree according to a maximum access count rule;

based on the value function of the each node in the Monte Carlo tree according to a maximum value function rule; or

based on the access count and the value function of the each node in the Monte Carlo tree according to a “maximum access count first, maximum value function next” rule.

14. A non-transitory computer-readable storage medium comprising a computer program or instructions which, when executed by a driving decision-making apparatus in an autonomous driving vehicle, cause the apparatus to perform operations comprising:

obtaining information of a current driving environment state of the autonomous driving vehicle, wherein the autonomous driving vehicle includes one or more sensors;

constructing a Monte Carlo tree based on the current driving environment state, wherein the Monte Carlo tree comprises N nodes, each node represents a corresponding driving environment state, the N nodes comprise a root node and N−1 non-root nodes, the root node represents the current driving environment state, a first driving environment state represented by a first node is predicted by using a stochastic model of driving environments based on a second driving environment state represented by a parent node of the first node and based on a driving action, the driving action is determined by the parent node of the first node in a process of obtaining the first node through expansion, the first node is any node of the N−1 non-root nodes, and N is a positive integer greater than or equal to 2;

determining, in the Monte Carlo tree based on at least one of an access count or a value function of each node in the Monte Carlo tree, a node sequence, wherein the node sequence comprises a plurality of nodes that starts from the root node and ends at a leaf node;

in response to determining the node sequence, determining a driving action sequence of a plurality of future driving steps, wherein each future driving step in the driving action sequence comprises a driving action corresponding to each node comprised in the node sequence, and wherein the driving action sequence is used for driving decision-making;

autonomously driving the autonomous driving vehicle based on a first driving action in the driving action sequence;

obtaining an actual driving environment state after the first driving action is executed; and

updating the stochastic model of driving environments based on the current driving environment state, the first driving action, and the actual driving environment state, wherein

the access count of each node is determined based on access counts of subnodes of the each node and an initial access count of the each node, the value function of the each node is determined based on value functions of subnodes of the each node and an initial value function of the each node, the initial access count of the each node is 1, and the initial value function of the each node is determined based on a value function that matches the corresponding driving environment state represented by the each node.

15. The non-transitory computer-readable storage medium according to claim 14 , wherein that the first driving environment state represented by the first node is predicted by using the stochastic model of driving environments based on the second driving environment state represented by the parent node of the first node and based on the first driving action comprises:

predicting, through dropout-based forward propagation by using the stochastic model of driving environments, a probability distribution of a driving environment state after the first driving action is executed based on the second driving environment state represented by the parent node of the first node; and

obtaining the first driving environment state represented by the first node through sampling from the probability distribution.

16. The non-transitory computer-readable storage medium according to claim 14 , wherein that the initial value function of the node is determined based on the value function that matches the driving environment state represented by the node comprises:

selecting, from an episodic memory, a first quantity of target driving environment states that have a highest matching degree with the driving environment state represented by the node; and

determining the initial value function of the node based on value functions respectively corresponding to the first quantity of target driving environment states.

17. The non-transitory computer-readable storage medium according to claim 16 , wherein the operations further comprise:

when a driving episode ends, determining a cumulative reward return value corresponding to an actual driving environment state after each driving action in the driving episode is executed; and

updating the episodic memory by using, as a value function corresponding to the actual driving environment state, the cumulative reward return value corresponding to the actual driving environment state after each driving action is executed.

18. The non-transitory computer-readable storage medium according to claim 14 , wherein the node sequence is determined:

based on the access count of the each node in the Monte Carlo tree according to a maximum access count rule;

based on the value function of the each node in the Monte Carlo tree according to a maximum value function rule; or

based on the access count and the value function of the each node in the Monte Carlo tree according to a “maximum access count first, maximum value function next” rule.

Assignments (1)

Priority Claims (1)

Continuity (2)

References Cited (22)