Post collision analysis-based vehicle action optimization for autonomous driving vehicles

• US 10,549,749 B2
• Filed: 03/08/2018
• Issued: 02/04/2020
• Est. Priority Date: 03/08/2018
• Status: Active Grant

• Current Assignee: Baidu Online Network Technology (Beijing) Co., Ltd (Baidu Incorporated), Baidu USA LLC (Baidu Incorporated)
• Original Assignee: Baidu USA LLC (Baidu Incorporated)
• Inventors: Zhu, Fan, Kong, Qi, Jiang, Hui
• Primary Examiner(s): Chen, Shelley
• Application Number:
  US15/772,523
• Publication Number:
  US 20190276012A1
• Time in Patent Office: 698 Days
• Field of Search: None
• CPC Class Codes: B60W 30/085 Taking automatic action to adjust vehicle attitude in preparation for collision, e.g. braking for nose dropping
• CPC Class Codes: B60W 30/0953 the prediction being responsive to vehicle dynamic parameters
• CPC Class Codes: B60W 30/0956 the prediction being responsive to traffic or environmental parameters
• CPC Class Codes: B60W 30/10 Path keeping cruise control for automatically following a preceding vehicle B60W30/165
• CPC Class Codes: B60W 60/001 Planning or execution of driving tasks
• CPC Class Codes: G05D 1/0088 characterized by the autonomous decision making process, e.g. artificial intelligence, predefined behaviours using knowledge based models G06N5/00
• CPC Class Codes: G05D 1/0214 in accordance with safety or protection criteria, e.g. avoiding hazardous areas monitoring the location of vehicles within a certain area, e.g. forbidden or allowed areas, in traffic control systems for road vehicles G08G1/13

Abstract

It is determined an ADV is inevitable to collide with at least one of the objects surrounding the ADV based on the relative positions and speeds between the ADV and the objects in view of the physical dimensions of the ADV. An object may be another vehicle, a pedestrian, a bicycle, a static obstacle. For each of the objects, a collision cost between the ADV and the object is calculated using a set of one or more cost functions or cost algorithms. A collision cost represents an amount of potential damage if the collision is to occur. One of the objects is selected whose collision cost is the least amongst the objects. A path or trajectory is planned for the ADV leading or towards to the selected object to minimize the potential damage when the ADV potentially collide with the selected object compared to the remaining objects.

Claims

1. A computer-implemented method for operating an autonomous driving vehicle, the method comprising:

determining that an autonomous driving vehicle (ADV) will be inevitable to collide with at least one of a plurality of objects surrounding the ADV based on relative positions and speeds between the ADV and the objects in view of physical limitations of the ADV;

for each of the objects, calculating a collision cost between the ADV and the object using a set of one or more cost functions, wherein the collision cost represents a potential damage if the collision is to occur;

selecting one of the objects that has a least collision cost amongst the objects; and

planning a trajectory towards to selected object to minimize the potential damage when the ADV potentially collides with the selected object compared to remaining objects.

2. The method of claim 1, wherein calculating a collision cost between the ADV and an object comprises determining an object collision cost based on a type of the object.

3. The method of claim 2, wherein determining an object collision cost comprises:

determining a type of the object based on perception information of the object; and

performing a lookup operation in an object to collision cost (object/cost) mapping table to locate a mapping entry matching the type of the object to obtain a predetermined collision cost corresponding to the type of the object.

4. The method of claim 3, wherein the object/cost mapping table comprises a plurality of mapping entries, and wherein each mapping entry maps a particular type of objects to a preconfigured collision cost.

5. The method of claim 2, wherein calculating a collision cost between the ADV and an object further comprises calculating a speed collision cost based on a difference between a first speed of the ADV and a second speed of the object.

6. The method of claim 2, wherein calculating a collision cost between the ADV and an object further comprises calculating a direction collision cost based on a first directional vector of the ADV and a second directional vector of the object.

7. The method of claim 2, wherein calculating a collision cost between the ADV and an object further comprises calculating a hitting location collision cost based on a hitting location of the potential collision with the object with respect to a driver seat of the ADV.

8. A non-transitory machine-readable medium having instructions stored therein, which when executed by a processor, cause the processor to perform operations, the operations comprising

determining that an autonomous driving vehicle (ADV) will be inevitable to collide with at least one of a plurality of objects surrounding the ADV based on relative positions and speeds between the ADV and the objects in view of physical limitations of the ADV;

for each of the objects, calculating a collision cost between the ADV and the object using a set of one or more cost functions, wherein the collision cost represents a potential damage if the collision is to occur;

selecting one of the objects that has a least collision cost amongst the objects; and

planning a trajectory towards to selected object to minimize the potential damage when the ADV potentially collides with the selected object compared to remaining objects.

9. The machine-readable medium of claim 8, wherein calculating a collision cost between the ADV and an object comprises determining an object collision cost based on a type of the object.

10. The machine-readable medium of claim 9, wherein determining an object collision cost comprises:

determining a type of the object based on perception information of the object; and

performing a lookup operation in an object to collision cost (object/cost) mapping table to locate a mapping entry matching the type of the object to obtain a predetermined collision cost corresponding to the type of the object.

11. The machine-readable medium of claim 10, wherein the object/cost mapping table comprises a plurality of mapping entries, and wherein each mapping entry maps a particular type of objects to a preconfigured collision cost.

12. The machine-readable medium of claim 9, wherein calculating a collision cost between the ADV and an object further comprises calculating a speed collision cost based on a difference between a first speed of the ADV and a second speed of the object.

13. The machine-readable medium of claim 9, wherein calculating a collision cost between the ADV and an object further comprises calculating a direction collision cost based on a first directional vector of the ADV and a second directional vector of the object.

14. The machine-readable medium of claim 9, wherein calculating a collision cost between the ADV and an object further comprises calculating a hitting location collision cost based on a hitting location of the potential collision with the object with respect to a driver seat of the ADV.

15. A data processing system, comprising:

a processor, and

a memory coupled to the processor to store instructions, which when executed by the processor, cause the processor to perform operations, the operations including

determining that an autonomous driving vehicle (ADV) will be inevitable to collide with at least one of a plurality of objects surrounding the ADV based on relative positions and speeds between the ADV and the objects in view of physical limitations of the ADV,

for each of the objects, calculating a collision cost between the ADV and the object using a set of one or more cost functions, wherein the collision cost represents a potential damage if the collision is to occur,

selecting one of the objects that has a least collision cost amongst the objects, and

planning a trajectory towards to selected object to minimize the potential damage when the ADV potentially collides with the selected object compared to remaining objects.

16. The system of claim 15, wherein calculating a collision cost between the ADV and an object comprises determining an object collision cost based on a type of the object.

17. The system of claim 16, wherein determining an object collision cost comprises:

determining a type of the object based on perception information of the object; and

performing a lookup operation in an object to collision cost (object/cost) mapping table to locate a mapping entry matching the type of the object to obtain a predetermined collision cost corresponding to the type of the object.

18. The system of claim 17, wherein the object/cost mapping table comprises a plurality of mapping entries, and wherein each mapping entry maps a particular type of objects to a preconfigured collision cost.

19. The system of claim 16, wherein calculating a collision cost between the ADV and an object further comprises calculating a speed collision cost based on a difference between a first speed of the ADV and a second speed of the object.

20. The system of claim 16, wherein calculating a collision cost between the ADV and an object further comprises calculating a direction collision cost based on a first directional vector of the ADV and a second directional vector of the object.

21. The system of claim 16, wherein calculating a collision cost between the ADV and an object further comprises calculating a hitting location collision cost based on a hitting location of the potential collision with the object with respect to a driver seat of the ADV.