Text this: Noise-resistant and scalable collective preference learning via ranked voting in swarm robotics