In this work, a LOcation based RAN resource Management and Access (LOMA) map is designed for future radio access networks (RAN) to allocate radio resources in complex wireless environment, and to facilitate uplink/downlink data transmissions. Enabled by the AI and high-precise positioning techniques, the LOMA map can associate a set of radio resources and data transmission parameters (e.g, transmit power, MCS level) with a geographical location in the RAN area. Given the LOMA map, each user equipment (UE) or infrastructure associated with the RAN can directly determine the radio resources and parameters used for uplink/downlink data transmissions according to its location. An AI enabled LOMA map generation method are proposed to generate LOMA maps according to the statistical traffic and wireless environment data collected by UEs and infrastructures. A reinforcement learning (RL) based algorithm is further proposed in the LOMA map generation method to dynamically quantify the available radio resources according to the real-time data traffic and the performance of applied resource scheduling scheme. Case studies with numerical results are presented to show the benefits provided by LOMA map technique in terms of increasing resource sharing efficiency, reducing signal overheads in data transmissions, and enabling resource scheduling schemes with less computing cost.