The top to bottom chain of events in controlling robots may be summarized as:
- User Issues a voice command that is converted into text.
- The commands are converted into the relevant robot-executable ones.
- The processed commands are sent to the Robot Server for effecting their
execution.
- Status of execution is reported to the user.
Several methods to realize (a) have been proposed and implemented but most of them have
been found to be working satisfactorily only in speaker dependent domains. We assume that
the user in this robots’ world can use one such technique to implement a speech
recognition system that can be trained to understand his voice. On a Microsoft Windows
platform, for instance, the MS-SAPI or Dragonsoft are viable alternatives for realizing
such a system.
Event (b), viz. the process of converting text commands to a target robot language, is
a difficult issue especially when each robot comprehends a different language. At the
first glance, this may look theoretically possible (!!) but several practical
issues make the problem a very complex one. Text commands have to be parsed to find out
the robot referred to, the task to be executed along with its nature (serial, parallel,
based on sensors, whether in real time, etc.) at a higher level of the language and then
converted to robot sensorimotor level commands that eventually effect the actuation
of the related motors based on the sensory conditions (event (c)). A reverse chain may be
followed to deliver the status of execution in voice output to realize event (d).
The embedded robots architecture has some more terminology and transport techniques.
After the event (a) at the Client machine, the text is sent a Central Server which may
either host the Natural Language Processor or send it to one that does so to realize (b).
The results of (b) are scheduled along with other commands issued to the same robot and
sent for final execution to the Server that hosts the robot. The last Robot Server feeds
the information to the relevant robot (more than one robot could be connected to a
Server), collects and sends back the status to achieve (c) and (d) via the Central Server.
back
