Estimation of a radio receiver's location from single-antenna observations - the received signal strength - is well known for its limited accuracy and lack of robustness. Yet, for reasons of energy- and space efficiency, emerging IoT devices will often be equipped with a single antenna. In this paper, we show how reconfigurable intelligent surfaces (RISs) can bring robustness and precision to this estimation problem. We propose a novel RIS-assisted SISO location scheme, based on new dynamic RIS reconfiguration protocols and an associated Maximum Like-lihood location estimation. We derive the Fisher information, the Cramér- Rao bound, and evaluate through simulations the effects of various relative RIS geometries and RIS reconfiguration pro-tocols. Our results indicate that the deployment of multiple RISs in the far-field allows for centimeter-level estimator accuracy. Reconfiguring RISs in a (pseudo-) random manner outperforms a deterministic orderly protocol by about 4 dB.