The emergence of cognitive radios and their opportunistic use of the radio spectrum calls for efficient spectrum sensing algorithms, able to reliably detect signals in very low Signal-to-Noise Ratio (SNR) environments with unknown or changing statistical distributions. Cyclic-feature detectors, which make use of the cyclostationary signature of communication signals, are a promising lead to meet these requirements. This paper presents a new nonparametric cyclic-feature detector based on the Spatial Sign Cyclic Polyspectrum (SSCP) that detects single-carrier signals for the aforementioned scenarios. In addition to being nonparametric the new detector is also robust with respect to narrowband interferers.