Modular and reconfigurable robotic systems have been designed to provide a\ncustomized solution for the non-repetitive tasks to be performed in a\nconstrained environment. Customized solutions are normally extracted from\ntask-based optimization of the possible manipulator configurations but the\nsolution are not integrated, for providing the modular compositions directly.\nIn this work, in the first phase, a strategy of finding unconventional optimal\nconfigurations with minimal number of degrees-of-freedom are discussed based\nupon the prescribed working locations and the cluttered environment. Then, in\nthe second phase, design of the modular and reconfigurable architecture is\npresented which can adapt these unconventional robotic parameters. Rather than\ngenerating and evolving the modular compositions, a strategy is presented\nthrough which the unconventional optimal configurations can be mapped directly\nto the modular compositions. The generated modular composition is validated\nusing Robot Operating System for the motion planning between the prescribed\nworking locations in a given cluttered environment.\n