Abstract: ‍ ‍The emerging technology of reconfigurable intelligent surface （RIS）， as one of the potential enabling key technologies for 6G， offers the ability to actively enhance and sense wireless propagation environments by intelligently manipulating the reflection characteristics of wireless signals through a large number of unit cells. Its application in angle perception and estimation has garnered widespread attention. However， in previous studies of hardware implementations of RIS for direction of arrival estimation， the unit cells have usually been controlled in rows or columns， which limits the spatial dimension degrees of freedom， limiting it to one-dimensional direction of arrival estimation. In this context， this paper presents a flexible RIS control platform designed to achieve independent and high-speed control of each unit cell， fully leveraging the spatial dimension of RIS hardware. Building upon this platform， this paper describes the realization of the real-time estimation of two-dimensional direction of arrival. First， we introduce the scheme for the RIS-based direction of arrival estimation system. By applying high-speed space-time coding modulation to the RIS， a mapping relationship between the harmonic components of the reflected signal and the direction of arrival of the incident signal is established. At the receiver， the amplitude and phase information of the harmonic components of the reflected signal are utilized to estimate the direction of arrival of the incident signal in real time. Next， based on the proposed scheme， we designed the wireless frame structure and key parameters of the system， constructing a prototype of the RIS-based two-dimensional direction of arrival estimation system. Finally， using the prototype， measurement experiments were conducted within the range of azimuth angle from -50 to 50 degrees and elevation angles from -30 to 40 degrees. Thirty-two measurement points were selected to estimate and record the azimuth and elevation angle of the incident signal. By analyzing the experimental data， it was found that 91% of the azimuth angle estimation results had an error of less than 5 degrees， whereas 78% of the elevation angle estimation results had an error of less than 5 degrees. The experimental results validated the feasibility of the proposed scheme. In the future， the system’s estimation accuracy can be further improved by optimizing the design of the RIS hardware and the space-time-coding matrices. Additionally， integrating the RIS-based direction of arrival estimation system with existing communication systems can be explored to achieve integrated sensing and communication， which can enhance wireless networks by enabling high-quality communication and high-precision perception simultaneously， thereby improving overall wireless network performance.