DeComplex: Task planning from complex natural instructions by a collocating robot

As the number of robots in our daily surroundings like home, office,\nrestaurants, factory floors, etc. are increasing rapidly, the development of\nnatural human-robot interaction mechanism becomes more vital as it dictates the\nusability and acceptability of the robots. One of the valued features of such a\ncohabitant robot is that it performs tasks that are instructed in natural\nlanguage. However, it is not trivial to execute the human intended tasks as\nnatural language expressions can have large linguistic variations. Existing\nworks assume either single task instruction is given to the robot at a time or\nthere are multiple independent tasks in an instruction. However, complex task\ninstructions composed of multiple inter-dependent tasks are not handled\nefficiently in the literature. There can be ordering dependency among the\ntasks, i.e., the tasks have to be executed in a certain order or there can be\nexecution dependency, i.e., input parameter or execution of a task depends on\nthe outcome of another task. Understanding such dependencies in a complex\ninstruction is not trivial if an unconstrained natural language is allowed. In\nthis work, we propose a method to find the intended order of execution of\nmultiple inter-dependent tasks given in natural language instruction. Based on\nour experiment, we show that our system is very accurate in generating a viable\nexecution plan from a complex instruction.\n