PAST WORK :

NATURAL LANGUAGE COMMUNICATION WITH MOBILE ROBOT :

    Mark C Torrance. AI Group . MIT Artificial Intelligence Laboratory.
Torrence implemented a system which mediated between an reactive mobile robot and domain restricted natural language. He introduced ROB's (Reactive odometric Plan) for plan recognition and fault tolerant plan execution and learning to associate human terms with location in the environment..
ROB : A reactive odometric plan or ROP is a representation of a short range plan to get the robot between two places. These ROP's provide a way for the robot to understand a place in terms of a way to get there in the world.
 
 
Plan (Rop *)
Rop (Starting place,Goal,(Step*))
Step (Heading,Length,Stop-condition)
Heading 0.0 .. 360.0 | nil
Length 0 .... C | nil
Stop Condition shortfront | longfront |shortleft | longleft | shortright |longright

The basic execution sequence of the robot consists of a loop. in which data is grabbed from the robot; natural language commands are processed and current ROP is executed to update robot's current location based on plan recognition.Robot replans as need to get it's goal.
 

TESTING A SIMULATED ROBOT THAT FOLLOW DIRECTIONS :

    Andre R Haas, Dept Of Com Science, Univ. Of Albany,
Dr. Haas's simulated robot travels through an office building by following directions in plain English. Although the robot is simulated ,the robotic high level procedures used to control it and it's capabilities are very realistic. All these procedures can be implemented using the very primitive senses and actions of a robot so this simulated robot is very close to the real one.
He used  grammar inspired by HPSG (Head Driven Phase Structured Grammar). There are few very general phrase structure rules which cut across boundaries different parts of speech. Each phrase structure rule contains semantics as well as syntax. Semantics derivation of his parser is very much close to the semantics rule of HPSG i.e.- Rule for subject and complement is that semantic representation of mother is equal to the semantic representation of the it's head daughter. The modifier also follows HPSG. for conjunctions rule is that if the representation of mother is the concatenation of the representation of daughters.
Rules for phrase structure are very simple leaving most of the semantics in the lexicon. So as the sentence is parsed and syntex analysis is done, semantic representation of the whole sentence is constructed simultaneously.