People Find-and-Follow Behavior for Service Robots using Continuous Belief Maps
Abstract
Find-and-follow is an important behavior for social robots which assist people. We introduce an on-line planning method for the robot: Continuous Real-time POMCP (CR-POMCP). It uses Partially Observable Monte-Carlo Planning (POMCP), which, in contrast to most other planning algorithms, can plan under uncertainty, on large state spaces, and real-time. Our method, CR-POMCP, also works in continuous state spaces, and moreover takes into account sensory noise, false negatives, and false positives.
Comparisons have been done with a Heuristic Follower, which simply follows the person. Two extensions to CR-POMCP were tested to improve the robot's performance when doing real-life experiments. All variants tested in simulation were found to be working better than the Heuristic Follower. Dynamic obstacles were introduced in simulation by adding randomly walking people. Real-life experiments have been done during a week with a mobile service robot in three urban environments of Barcelona with other people walking around.
Experimental Videos
Videos demonstrating the find-and-follow experiments. The robot searches and follows a person recognized using AR Markers. More details about the used urban environments are available on the maps page.
Video Legend
Each experiment video shows three areas:
- Video: Shows the scenario
- Map: Map as shown by ROS rviz with:
- Dabo: Blue body, white head
- Obstacles: Black and dark gray
- Laser detections: White line/dots
- People detection:
- Leg detection: White dots
- AR Marker detection: Blue dot
- Last used person location: Red dot (combination of leg and AR Marker detection)
- Belief Map: Shows the robot's belief (probability matrix) of the person's location:
- Dabo: Blue circle
- Obstacles: Black squares
- Probability matrix: Light blue (0) to white (low) to red (high probability)
Experiments
Heuristic Follower vs Adaptive HB-CR-POMCP Follower
The Heuristic Follower goes to the last position where it saw the person and waits until it sees the person again. In contrast, the Adaptive HB-CR-POMCP Follower maintains a belief (memory) with the location of the person. The video shows two experiments on the FME lab, where the Heuristic Follower does not move until the person is visible, while the Adaptive HB-CR-POMCP Follower actively moves to find the person.
Adaptive HB-CR-POMCP Follower in the BCN Lab
The method works in large urban environments. The video shows experiments on the Barcelona Robot Lab, including an example where the robot follows the highest belief point when not seeing the person, and the belief stays consistent with the person's movement.
Find
This video demonstrates the method's ability to find the person, even if they have not yet been seen before.
Group
The method also works when other people are present and obstructing the robot's vision.
Long Run in the Barcelona Robot Lab
The person is followed in the Barcelona Robot Lab over an extended period.
Long Run at Telecos Square
The person is followed in another urban environment, the Telecos Square on the campus.
If you cannot see the complete video, please right click it and choose the option "Copy URL" or "Copy Location", and then try to view it with an external viewer (e.g. VLC) or another browser.
More Information
More details about the used urban environments are available on the maps page, where they can also be downloaded.