Distribution of goods is one of the major aspects of supply chains. At first, manufacturers used the point-to-point approach for sending their goods. Then, it became required for having a liaison between manufacturers and consumers. Warehouses had a significant role in supply chains by stocking and sending goods in appropriate time. Having goods with high inventory costs or goods with short delivery time resulted in a new approach named cross docking. Cross docks decreased inventory and ordering costs. Therefore, many researchers and business owners paid attention to cross docks. Recent studies have shown that such storehouses can be used efficiently for gathering and recycling returned goods along with using in distribution of commodities. The lack of research for integration of forward and backward logistics has caused a gap between scientific research and the real world. This study aims to bridge the gap by introducing a model. The proposed model has the following assumptions: (a) A vehicle cannot visit a point more than once in each period. (b) Open routing of vehicles in receiving and recycling parts. (c) Vehicle routing with pickup and delivery for receiving and recycling in the delivery part. (d) The problem is a type of multi-period and multi-product model. Due to the aforementioned discussion, a linear model is presented, and some problems are solved. Small and moderate-sized problems were solved by CPLEX solver in GAMS in less than one hour. Because this model is a kind of NP-hard problems, the variable neighborhood search (VNS) has been presented for solving large-sized problems. By comparing exact solutions with results of the presented metaheuristic algorithm, little amount of error was found, and the efficiency of the algorithm was high. At the end, the presented model has been applied to a case study which consists of two baguette production workshops, 44 fast foods, and a recycling center. Routing is also performed for baguette distribution and collecting returned baguettes.