Path determination for autonomous vehicle parking

Patent metadata

Field	Value
Publication number	US-12589768-B2
Application number	US-202418590375-A
Country	US
Kind code	B2
Filing date	Feb 28, 2024
Priority date	Feb 28, 2024
Publication date	Mar 31, 2026
Grant date	Mar 31, 2026

Abstract

A method of determining a movement path for a vehicle includes determining a set of adjacent nodes with respect to a current node. The method includes determining, via a neural network, a subsequent path node and determining whether a path based on the subsequent path node connects a source to a goal node. The method includes, in response to a determination that the path based on the subsequent path node does not connect the source node to the goal node, determining an alternative subsequent path node, and determining, via the neural network, a second path based on the alternative subsequent path node. The method includes, in response to a determination that the path based on the subsequent path node connects the source to the goal node, selecting the path based on the subsequent path node as the movement path, and executing a set of actions associated with the movement path.

First claim

The invention claimed is:

1 . A method of determining a movement path for a vehicle from a source node to a goal node, the source node corresponding to a current position and a current orientation of the vehicle, the goal node corresponding to a goal position and a goal orientation of the vehicle, and the goal node being defined with respect to the source node, the method comprising:
determining a set of adjacent nodes, wherein each node of the set of adjacent nodes represents movement of the vehicle with respect to a current node; determining a set of costs associated with the set of adjacent nodes; determining a path-starting node based on a current node by:
determining, via a neural network, a predicted node based on the current node, wherein the predicted node represents movement of the vehicle with respect to the current node; and
selecting the predicted node as the path-starting node;
determining, via the neural network, a first set of path nodes based on the path-starting node; determining whether a path defined by the first set of path nodes connects the source node to the goal node; in response to a determination that the path defined by the first set of path nodes does not connect the source node to the goal node:
determining an alternative path-starting node based on a lowest cost node of the set of adjacent nodes, and
determining, via the neural network, a second set of path nodes based on the alternative path-starting node, and
in response to a determination that the path defined by the first set of path nodes connects the source node to the goal node:
selecting the first set of path nodes as the movement path, and
executing a set of actions associated with the movement path.

2 . The method of claim 1 wherein the set of actions includes autonomously moving the vehicle along the movement path.

3 . The method of claim 1 wherein the set of actions includes automatically controlling steering of the vehicle along the movement path.

4 . The method of claim 1 wherein the set of actions includes automatically controlling steering, acceleration, and braking of the vehicle along the movement path.

5 . The method of claim 1 wherein the set of actions includes displaying prompts to a driver for moving the vehicle along the movement path.

6 . The method of claim 1 wherein the goal node corresponds to a vehicle parking position.

7 . The method of claim 6 further comprising receiving location and orientation information for the goal node from a parking spot selection module.

8 . The method of claim 1 further comprising, in response to a determination that the path defined by the second set of path nodes connects the source node to the goal node:
selecting the second set of path nodes as the movement path, and
executing a second set of actions associated with the movement path.

9 . The method of claim 1 wherein the neural network is generated via a machine learning model trained by reinforcement learning via repeated simulations of vehicle movement in varied environments.

10 . The method of claim 1 wherein:
a respective cost of the set of costs is associated with a respective node; and
the respective cost of the respective node is based on:
a quantity of movement direction changes associated with the respective node,
a quantity of movement direction changes associated with a path formed from the source node to the respective node,
a path length associated with the path formed from the source node to the respective node,
a magnitude of a steering angle associated with the respective node,
a magnitude of change in a steering angle from a previous node to the steering angle associated with the respective node, and
whether the respective node was recommended by the neural network.

11 . The method of claim 10 wherein the path length is measured in terms of at least one of:
a number of nodes, and
a total movement distance.

12 . The method of claim 1 wherein a respective node of the set of adjacent nodes is associated with a set of coordinates including:
an x-coordinate corresponding to a possible vehicle location;
a y-coordinate corresponding to the possible vehicle location; and
an angle corresponding to a possible vehicle orientation at the possible vehicle location.

13 . The method of claim 12 wherein the respective node of the set of adjacent nodes is associated with a first cost based on the possible vehicle location, a first movement distance, a first movement direction, and a first steering angle required to reach the possible vehicle location from the current node.

14 . The method of claim 13 further comprising:
determining, based on a cost associated with the current node, whether the respective node of the set of adjacent nodes is associated with a cost above a cost threshold; and
in response to the determination that the respective node is associated with a cost above the cost threshold, excluding the respective node from the set of adjacent nodes.

15 . The method of claim 1 further comprising receiving, from an obstacle detection module, data corresponding to a set of obstacles.

16 . A system of determining a movement path for a vehicle from a source node to a goal node, the source node corresponding to a current position and a current orientation of the vehicle, the goal node corresponding to a goal position and a goal orientation of the vehicle, and the goal node being defined with respect to the source node, the system comprising:
memory hardware; and processor hardware communicatively coupled to the memory hardware, wherein the processor hardware is configured to:
determine a set of adjacent nodes, wherein each node of the set of adjacent nodes represents movement of the vehicle with respect to a current node;
determine a set of costs associated with the set of adjacent nodes;
determine a path-starting node based on a current node by:
determining, via a neural network, a predicted node based on the current node, wherein the predicted node represents movement of the vehicle with respect to the current node; and
selecting the predicted node as the path-starting node;

determine, via the neural network, a first set of path nodes based on the path-starting node;
determine whether a path defined by the first set of path nodes connects the source node to the goal node;
in response to a determination that the path defined by the first set of path nodes does not connect the source node to the goal node:
determine an alternative path-starting node based on a lowest cost node of the set of adjacent nodes, and
determine, via the neural network, a second set of path nodes based on the alternative subsequent path-starting node, and

in response to a determination that the path defined by the first set of path nodes connects the source node to the goal node:
select the first set of path nodes as the movement path, and
execute a set of actions associated with the movement path.

17 . The system of claim 16 wherein the processor hardware is configured to, in response to a determination that the path defined by the second set of path nodes connects the source node to the goal node:
select the second set of path nodes at the movement path, and
execute a second set of actions associated with the movement path.

18 . The system of claim 16 wherein:
a respective cost of the set of costs is associated with a respective node;
the respective cost of the respective node is based on:
a quantity of movement direction changes associated with the r

Assignees

• Aptiv Technologies AG

Inventors

• Sugiarto Tanto
• Orlowski Mateusz
• Turlej Wojciech

Classifications

• B60W2552/50

  Barriers · CPC title

• B60W60/001

  Planning or execution of driving tasks · CPC title

• B60W30/08

  Active safety systems} predicting or avoiding probable or impending collision {or attempting to minimise its consequences · CPC title

• B62D15/0285 Primary

  Parking performed automatically · CPC title

• B60W30/06 Primary

  Automatic manoeuvring for parking · CPC title