In this paper, a method for improving wireless localization of active targets in indoor environments is proposed. The positioning problem is recast to an inverse problem and solved by means of multi-objective evolutionary optimization. The first objective considers received signal strength wireless positioning based on a theoretical propagation model. The second objective relies on probabilistic information that takes into account a-priori knowledge about the environment and semantic profile inferred from the target. The system has been implemented and experimentally validated in a real scenario using smart-phones and the existing Wi-Fi network infrastructure.