Faster Exact Methods for Large Scale Real-time Logistics Operations — Implications for Practice

Problems arising in warehousing and distribution continue to grow in size and complexity. In addition, we may finally be at an inflexion point in which automation will finally impact all corners of large scale logistics operations. While many metaheuristics or simpler techniques can be helpful in solving the many combinatorial optimization problems that arise, our progress in improving the speed of column generation (exact methods) for solving large scale integer linear programs means that these methods will soon be practical for problems such as capacitated vehicle routing (with and without time windows or deadlines), robot routing (with and without human handlers) in single- and multi-level warehouses and product placement problems which are a key to effective large-scale fulfillment operations. This talk will present a number of problems, some details about our new pre-computing methods that result in remarkable (10-100X) speedups and our plans for implementation of these methods in practice.