Nature is considered one promising area to search for inspiration in designing robotic systems. Some work in swarm robotics has tried to build systems that resemble distributed biological systems and inherit biology's fault tolerance, scalability, dependability, and robustness. Such systems, as well as ones in the areas of active self-assembly and amorphous computing, typically use relatively simple components with limited computation, memory, and computational power to accomplish complex tasks, such as forming paths in the presence of obstacles. We demonstrate that such tasks can be accomplished in the well-studied tile assembly model, a model of molecular self-assembly that is strictly simpler than other biologically-inspired models. Our systems use a small number of distinct components to find minimal-length paths in time linear in the length of the path while inheriting scalability and fault tolerance of the underlying natural process of self-assembly.