Human-machine interface with visual communication
When you make a mistake in the input of computer command,
the computer may say "no", even though the mistake is trivial and
it is easely recovered.
If the computer recognize the sequence of the input commands and
understand what its user is doing, there may be some way to estimate
the improper input.
On this assumption, we are developing a multimodal interface which
examines the context of user's input, estimates the probable input
form and correct mistakes if any.
We adopt two ways: one uses huristics, and the other uses logical
reasoning about human belief and predicts what is facts.
We are developing a multi-agent system which understands
qualitative representation of surrounding space including moving
objects and communicates with us.
Some of Papers
- Kenji Kanai, Takashi Watanabe,
High Accuracy Calibration of Multiple Cameras by TAGUCHI Method and
Multiple Viewpoints Tracking of Human Motion Flows, Flow
Visualization, 6A-3, (2010).
- Hiroyuki Furukawa, Takuya Takagishi, Takashi Watanabe,
Interactive Visualization System of Flow Development in
Taylor Vortex Flow, Flow Visualization, 4.2, (2006).
- Daisuke Date, Takashi Watanabe,
Dynamic Calibration of Image Sensors for the Observation and
Measurement of Flows, Flow Visualization, 4.8.2:80, (2004).
Back to Research List.