Obtaining people's position information is essential for performing robotic services toward people. Imagine a robotic service that a robot helps somebody who lost his/her way. The robot first needs to find out somebody lost and wandering around in the crowded situation, obtains the position of the person, and then approach that person in trouble. Also, knowing what kind of place the person in target resides (semantic information of the space: spatial primitives), or how the person is behaving (semantic information of the behavior: behavioral primitives), will be helpful in deciding what service the robot shall provide or how the robot shall behave.
The goal of this project is to construct a common platform system from which robots can retrieve position or semantic information of the person in the environment. Such a common system will enable developers to easily create varieties of new robotic services, and will evenually help robots spread in homes and townsides. In order to appoach this goal, we are developing methods for people positioninig, and for extracting semantical information from these positions. Also, a testbed system will be constructed which relizese a common platform system providing these functionalities.
- Enable tracking of 5 to 10 persons in indoor environment, and obtain each person's position. Positioning error be less than 10cm for 90% case (mean error within 5 cm).
- For indoor environments with more than 10 person, enable positioning of persons equipped with ID tags, in a 3m mean error.
- With outdoor positioning by GPS and by integrating these results, realize a seamless people positioning throughout indoor and outdoor environments.
- Environmental information structuring:
- Develop a mean to extract primitives from 1 week storage of positioning data.
- Define some standard spatial primitives and behavioral primitives necessary for field trial experiments.
- Develop a common platform language that can describe various primitives and robotic services.
- Common platform system:
- Construct and manage a testbed system at Keihanna Info-Communication Openlab.
- Release the common platform softwards to public (free for research purpose only), such that robotic service developers can utilize them.