A New Social Robot Cooperation Model to Localize People in an Urban Environment
Abstract
We present a novel method to localize people in urban settings using cooperative social robots, that overcomes the limitations of existing approaches, which are either tailored to tightly bounded environments, or based on unrealistic human behaviors. Using this cooperative method, robots are able to find people who are beyond the reach of the robot's sensors. This approach includes people search, tracking, multi-robot cooperation and communication.
In particular, we define a "Cooperative Highest-Belief Continuous Real-time POMCP" method, which is able to run in real-time and in large continuous environments. This method makes use of the online search algorithm Partially Observable Monte-Carlo Planning (POMCP), which, in contrast to other previous approaches, can plan under uncertainty on large state spaces. The search strategy's decision makes a balance between the probability of a person being at a certain location, the distance to the locations, and if the location is close to the search goal already assigned to another robot. The validation of the model is accomplished throughout an extensive set of simulations and real-life experiments with one person and two robots.
Experimental Videos
Videos demonstrating the experiments done with the Cooperative HB-CR-POMCP to find and follow people. The robots search and follow a person recognized using AR Markers.
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
- Tibi: Orange body, white head
- Obstacles: Black and dark gray
- Laser detections: Blue/orange line/dots (from Dabo and Tibi respectively)
- Path: Blue/orange lines showing executed path
- People detection:
- Leg detection: Blue/orange dots
- Last used person location: Red dot (combination of leg and AR Marker detection)
- Belief Maps: Shows each robot's belief (probability matrix) of the person's location (Tibi top right, Dabo bottom right):
- Robot self: Large blue circle
- Other robot: Small blue circle
- Detected person's location: Large red circle
- Detected person's location by other robot: Small red circle
- Obstacles: Black squares
- Probability matrix: Light blue (0) to white (low) to red (high probability)
Experiments
Searching
The robots search cooperatively for the person by each looking at different places.
Following the Person in the FME Lab
The person is being followed by two robots, which goes very well. Even though the person should be visible, they are not always detected due to false negative detections. In these cases the detection of the other robot can be used, and otherwise the belief. In some situations the robots have difficulties navigating, which is due to the dimensions of the robots and the relatively small space.
Difficult Search
In this example the person hides very well, and the robots do not find them initially. Therefore the robots go to the other possible hiding location. Nevertheless, after having explored the second location, the robots return to the first place and find the person.
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.