Humanoid Robot Presentation Through Multimodal
Presentation Markup Language
Author
Doshi Pratik H.
Introduction♦ Not only documents but also multimodal contents including
visual and auditory information are used in media and interface fields these days.
♦ In order to make multimodal contents using lifelike characters easily, several languages have been developed [1].
♦ As MPML is a mediumlevel scripting language, anyone can write multimodal presentation with lifelike characters.
♦ Humanoid Robot are expected to play an important role in all types of fields.
♦ The presentation by the humanoid robot gives a di� erent impression, as more life like.
♦ Here MPMLHR uses humanoid robots as presenter, we can extend presentation contents explaining actual objects
MULTIMODAL PRESENTATION MARKUP LANGUAGE MPMLA
Features of MPML♦ Independent of particular systems♦ Character control function♦ Easy description♦ Media synchronization♦ Metadata description
MPML family♦ There are several versions of MPML in the MPML family. ♦ MPML Ver.1.0 is the first version developed. ♦ MPML Ver.2.0e [7], is an extension of Ver.2.0.♦ In MPML Ver.3.0, some extensions including a graphical
editor have been incorporated. ♦ MPMLVR [8] stands for MPML for Virtual Reality.♦ MPMLHR is designed based on this Ver.2.0e.
Figure 2: Specification tree of MPMLHR
MPML tags
A SYSTEM OF MPMLHR PRESENTATION
A
IS
OM
System Configuration♦ Humanoid robot ASIMO♦ Control Servers♦ ASIMO Voice Server♦ Receive Server♦ MPMLHR parser♦ Browser♦ LCD projector ♦ Speaker ♦ Vertical large screen
Figure 7: System configuration
MPMLHR Parser
Figure 8: MPMLHR control diagram
..2 MPMLHR Parser♦Opens the MPMLHR source code♦Examines it with its XML format.♦XSLT processor converts source code to operation data.♦It is understandable for the Receive Server.♦The Receive Server is connected with the ASIMO Voice
Server, Control Servers and the Browser.♦The Receive Server interprets the data into control commands
referring the mapping table and sends the commands through the TCP/IP connection to the ASIMO control.
Making Presentation More Lifelike♦Parallel execution of the action command and the speech
command
…2Making Presentation More Lifelike♦2 threads are generated in the Receive Server.♦The Receive Server receives the operation data from the
MPMLHR parser, and dispatches the commands to the ASIMO Voice Server or Control Servers or Browser.
♦If action thread is in idle state when Receive Server receives action command, the action command is sent to the Control Servers immediately to check for state of speech thread.
♦ When the Receive Server sends an action/speech command to the Control/voice Server, the action/speech thread turns into busy state from idle state until the end signal is received from the Control Servers.
♦If the action thread is in busy, the action command is sent to the Control Servers after both thread come in idle state.
…3Making Presentation More Lifelike♦Autonomous head motion
♦Action thread is in idle state while speaking.♦ To make more lifelike, the head action command is automatically sent at random.♦ The head action command makes the robot move his head right & left, and up & down. ♦It is sent at the time when the robot is speaking with no action.
…4Making Presentation More Lifelike♦This happens only when speech thread is in busy state and action thread is in idle state. ♦There are two situations in receiving the head action command, as shown in the figure. ♦One is that the robot is speaking with no action. ♦The other is, if the robot is speaking with the short action which ends before the speech ends. Tim ing of autonom ous head m otion
execution
Conclusions:♦MPMLHR is an extension of MPML.♦MPMLHR is very simple language♦Anyone can write attractive multimodal presentation
modules using a humanoid robot without expertise.♦The Receive Server has 2 threads to allow parallel execution
of the action command and the speech command making the robot presentation more lifelike.
♦It gives stronger impression looking at the humanoid robot presentation.