The design of cleaning and maintenance (CaM) robots is generally limited by their fixed morphologies, resulting in limited functions and modes of operation. Contrary to fixed shape robots, the design of reconfigurable robots presents unique challenges in designing their system, subsystems, and functionalities with the scope for innovative operational scenarios and achieving high performance in multiple modalities without compromise. This paper proposes a heuristic framework using three layers, namely input, formulation, and output layer, for designing reconfigurable robots with the aid of established transformation principles including expand/collapse, expose/cover, and fuse/divide observed in several products, services, and systems. We apply this heuristic framework approach to the novel design of a pavement CaM robotic system and subsystems, namely, (i) varying footprint, (ii) transmission, (iii) outer skin or cover, (iv) storage bin, (v) surface cleaning, and (vi) vacuum/suction and blowing. The advances in the design method using the heuristic approach are demonstrated by developing an innovative reconfigurable design for the CaM task. Kinematic analysis and control architecture enables the unique locomotion behavior and gaits, namely, (a) static reconfiguration and (b) reconfiguration while locomotion, supported by the control architecture. Experiments were conducted, and outcomes were discussed along with the failure mode analysis to support the design robustness and limitations through the observations made from the development to testing phase over one year. A detailed video demonstrating the design capabilities is linked.