This paper proposes a novel data clustering algorithm, coined ?cellular ants?, which combines principles of cellular automata and ant colony optimization algorithms to group similar multi-dimensional data objects within a two-dimensional grid. The proposed method assigns data objects to unique ants, which actively move around, leave pheromones and follow trails of similar ants. Cellular automata principles based on simple, discrete neighborhood densities determine an ant?s directional movements, so that clusters emerge. The novel concept of ?positional swapping? organizes these clusters internally based on multi-dimensional data value similarity. As a result, shared cluster borders in grid space contain data objects that are nearby in parameter space. This method is algorithmically simple, as it is based on a few user-chosen variables and uses fixed discrete values instead of probability algorithms. This clustering technique is evaluated using several datasets, while its methodology and computational performance is compared to similar approaches.