Molecular robots are nanoscale devices that take input from the environment (senses) and process the information (computes) to inform an actuation. Despite a multitude of interesting molecular robots made of DNA, implementing dynamic state information is a major challenge. This thesis aims to leverage computational model-driven design principles for reconfigurable molecular robotics. We study two theoretical models consisting of monomer units that are globally connected and encode simple state information: Turning Machines utilise integer state information to control local rotations of robotic units to fold into target shapes. Despite the intended simplicity of the model, the state information and characteristic motion proved conceptually challenging to implement experimentally. We then propose the Tethered Tile Bot model, which consists of a chain of ‘tethered’ square tiles that encode target bonds as ‘glues’ on tile edges and use two rule types for reconfiguration: ‘bond break’ and ‘bond form’ rules. We implement Tethered Tile Bots using double-layered DNA origami tiles that are tethered to each other by a long single-strand of 360 DNA bases. Reconfiguration is realised as a two-step process. First, bonds between tiles are broken via toehold-mediated strand displacement. Then, bonds for the target configuration are formed via hybridisation. The tethers between tiles ensure the tiles do not fully dissociate into the liquid environment, allowing the original material to be reconfigured. Tethered Tile Bot configurations correspond to paths in the plane Z2, represented by a series of forward, left and right moves. Lastly, we implement a form of memory that encodes the target shape during the anneal, via custom designed sequences (memory domains). When precisely the same ‘information’ (DNA strands) is added to samples with different target shapes encoded, we observe different behaviours (target configurations) during the reconfiguration process. This serves as a step towards using computation to drive robotic molecular reconfiguration.