Robotic simulation using natural language commands

Robots are inevitable these days,so naive users should not find difficult to interact with robots. Since robots understand only RCL (Robot command language), we need a system which converts natural language commands into RCL. We use a semantic parser to address this problem of converting natural language commands to RCL that can be readily implemented in a robot execution system. Our system gets the natural language command from the user and converts it into RCL using tagging approach. This tagging operation is implemented using a trainer, which uses Hidden Markov Model approach. Using this tagged command the Parser builds the RCL. Then the RCL is converted to configurations which is the co-ordinates of the objects in a given spatial context. The validation of these configurations is performed using robotic simulator. We have used an annotated dataset to compare and evaluate our approach. Despite the fixed domain, the task is challenging as correctly parsing commands requires understanding spatial context.