Visible light positioning (VLP) systems provide higher accuracy and lower cost, more suitable for indoor positioning compared to their radio-frequency (RF) counterparts. And the VLP system based on received signal strength (RSS) is widely used due to the lowest hardware cost. However, existing RSS-VLP systems are usually sensitive to reflections and shadows. In this paper, we propose a VLP framework with an adaptive algorithm module, a diversity module, and a combination module. Simulations show that in the presence of reflections and shadows, this positioning framework can achieve an average error below 15 cm in a 5 m × 5 m area. Moreover, in case of severe blocking, its positioning error is even an order of magnitude lower than that of the typical RSS-VLP method.