Crowd-driven mapping, localization and social-friendly navigation system
Disclosed are methods of navigating a robot in a crowd involving mapping a local crowd and applying a clustering algorithm that reconstructs a crowd-flow map capturing movement patterns of pedestrians in the local crowd; using a flow-matching metric, localizing the robot in the crowd-flow map; and following a movement pattern in the crowd-flow map using a hierarchical crowd-driven planning on a long time horizon and a short time horizon.
1. A method of navigating a robot in a crowd, comprising:
mapping a local crowd and applying a clustering algorithm that reconstructs a crowd-flow map capturing movement patterns of pedestrians in the local crowd;
using a flow-matching metric, localizing the robot in the crowd-flow map; and
following a movement pattern in the crowd-flow map using a hierarchical crowd-driven planning on a long time horizon and a short time horizon.
2. The method according to claim 1 , wherein mapping a local crowd comprises detecting and tracking local pedestrians.
3. The method according to claim 1 , wherein mapping a local crowd comprises using a K-means clustering methodology to recover time-invariant crowd-flow patterns locally around the robot.
4. The method according to claim 1 , wherein localizing the robot in the crowd-flow map comprises using the flow-matching metric to evaluate a matching quality between a local crowd observation and the crowd-flow map locally around a given location.
5. The method according to claim 1 , wherein following a movement pattern comprises developing a resistance cost and a lubricating cost to encourage the robot to follow crowd-flow patterns.
6. The method according to claim 1 , wherein reconstructing the crowd-flow map comprises fusing locals map obtained in an online manner when the robot moves around in a local area.
7. The method according to claim 1 , wherein localizing the robot in the crowd-flow map using the flow-matching metric, further using a conventional geometry-matching metric, and then using the flow-matching metric and the conventional geometry-matching metric simultaneously to fuse the crowd and geometry information.
8. A robot navigating system, comprising:
a memory that stores computer executable components; and
a processor that executes the computer executable components stored in the memory, wherein the computer executable components comprise:
a crowdmapping module that maps a local crowd and applying a clustering algorithm that reconstructs a crowd-flow map capturing movement patterns of pedestrians in the local crowd;
a crowdlocalizer module that uses a flow-matching metric, localizing the robot in the crowd-flow map; and
a crowdplanner module that follows a movement pattern in the crowd-flow map using a hierarchical crowd-driven planning on a long time horizon and a short time horizon.
9. The robot navigating system according to claim 8 , wherein the crowdmapping module further detects and tracks local pedestrians.
10. The robot navigating system according to claim 8 , wherein the crowdmapping module further uses a K-means clustering methodology to recover time-invariant crowd-flow patterns locally around the robot.
11. The robot navigating system according to claim 8 , wherein crowdlocalizer module further uses the flow-matching metric to evaluate a matching quality between a local crowd observation and the crowd-flow map locally around a given location.
12. The robot navigating system according to claim 8 , wherein the crowdplanner module further develops a resistance cost and a lubricating cost to encourage the robot to follow crowd-flow patterns.
13. The robot navigating system according to claim 8 , wherein the crowdmapping module further comprises fusing locals map obtained in an online manner when the robot moves around in a local area.
14. The robot navigating system according to claim 8 , wherein the crowdlocalizer module further uses a conventional geometry-matching metric, and then uses the flow-matching metric and the conventional geometry-matching metric simultaneously to fuse the crowd and geometry information.
15. An autonomous vehicle comprising the robot navigating system according to claim 8 .
16. A delivery device comprising the robot navigating system according to claim 8 .
17. A non-transitory machine-readable storage medium, comprising executable instructions that, when executed by a processor, facilitate performance of operations, comprising:
mapping a local crowd and applying a clustering algorithm that reconstructs a crowd-flow map capturing movement patterns of pedestrians in the local crowd;
using a flow-matching metric, localizing the robot in the crowd-flow map; and
following a movement pattern in the crowd-flow map using a hierarchical crowd-driven planning on a long time horizon and a short time horizon.
18. The non-transitory machine-readable storage medium according to claim 17 , wherein the operations further comprise:
mapping a local crowd comprises using a K-means clustering methodology to recover time-invariant crowd-flow patterns locally around the robot.
19. The non-transitory machine-readable storage medium according to claim 17 , wherein the operations further comprise:
localizing the robot in the crowd-flow map comprises using the flow-matching metric to evaluate a matching quality between a local crowd observation and the crowd-flow map locally around a given location.
20. The non-transitory machine-readable storage medium according to claim 17 , wherein the operations further comprise:
following a movement pattern comprises developing a resistance cost and a lubricating cost to encourage the robot to follow crowd-flow patterns.