A robotic mobile fulfillment system (RMFS) is a “parts-to-picker” system employing a fleet of autonomous vehicles (AV), which transport pods between a storage area and picking stations. In this paper, an AV interference-free scheduling on bidirectional paths (IFSB) approach is studied and shows better working efficiency than unidirectional paths. In order to model AV scheduling on bidirectional path with interference-free constraints, a no-wait flexible process job shop scheduling problem (NWFPJSP) is employed to evaluate total completion time. A mathematical model aiming to minimize the total completion time is presented with as constraint interference-free operation. An A* algorithm is modified for path planning and a simulated annealing algorithm (SA) is employed for scheduling AVs and their paths. Based on a case study with a small RMFS, the IFSB approach results in a 40% increase of the storage capacity and a 22% decrease of the total completion time when compared to a unidirectional approach. Meanwhile, the number of AVs stop-starts are decreased by 40% and the paths length for all AV are decreased by 36%. This indicates the great potential of bidirectional paths in future logistic applications, which may increase the storage capacity, working efficiency, technical health of the AVs, and environmental sustainability.