Mobile Manipulation Hackathon (IROS 2018) - IRI Team
An IRI Team member from PALRobotics presents a restaurant service robot demonstration at the Mobile Manipulation Hackathon at IROS 2018.
In this talk
- 0:00
Serving a chosen dish
The demonstration frames food service as a choice between two dishes. Voice recognition worked in other tests, but the team says it was unavailable here, so a person tells the robot which dish to pick. The robot attempts to move a plate with food, though locating and handling it prove difficult.
- 1:16
Handling food and table limits
The team describes placing yogurt on a plate and notes this is their first attempt to pick up a loaded plate. The demo includes an attempt to serve the food. Table location is preprogrammed, and the presenter says navigation is not part of the setup, limiting how the robot can find things.
- 3:08
Asking before feeding
The robot looks for a spoon and has a programmed interaction for offering food. It asks the user before proceeding, uses a camera to check that the person is present, and starts feeding when it detects an open mouth. The presenter describes mouth detection as part of the scenario, while acknowledging that the system is not highly accurate.
- 7:50
Learning motions by demonstration
The team also wanted to show how it teaches the robot food handling motions. It learns trajectories from demonstrations and then generalizes them, with movement depending on where the blade is. The presenter contrasts this approach with a text-based setup that lacked the required teaching input.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.