O.R. Optimization Algorithms Drive Unmanned Warehouse Robots to Work


JD.com pioneered same-day delivery as a standard service in China’s B2C e-commerce sector 10 years ago. As both demand and labor costs have increased dramatically since then, the company has faced an emerging challenge of how to cost-effectively meet growing demands while maintaining high quality logistics services. In response, JD.com began to invest in unmanned warehouses in 2015.

One of the major challenges in an unmanned warehouse is the management and dispatching of automatic guided vehicles (AGVs). In a modern parts-to-picker automated warehouse system, AGVs pick up movable racks and transport inventory to workstations for picking.

A central question is how to effectively dispatch AGVs to racks and workstations in real time, to minimize the travel-distance and order fulfillment time. The problem can be formulated as a large-scale integer program with millions of decision variables and constraints. Moreover, it must be solved in real time. We tackle this problem via a Lagrangian relaxation together with cutting planes, and are able to find near optimal dispatching decisions for hundreds of AGVs in seconds.

Optimization combined with modern robotics technology results in tremendous improvements in picking efficiency and warehouse space utilization. JD.com currently operates more than 900 warehouses with an area of approximately 21 million square meters.

JD.com’s application of unmanned technologies is the most extensive in the field of e-commerce logistics in China. It has helped the company to lower the inventory turnover days to 33.3 despite the significant increase in the number of SKUs, and decrease fulfillment expense ratio to a world-leading level of 6.5%. In 2020, 90% of JD.com’s first-party retail orders were delivered the same- or next-day after they were placed.

The agility of such unmanned technologies not only makes it possible for a single warehouse to handle more than 1.3 million orders per day during peak sales seasons, but also helped the company quickly respond to COVID-19 and ensure the rapid recovery of production capabilities. Moreover, before the deployment of unmanned warehouses, logistics workers across the industry had faced unprecedented pressure in the form of physical labor. For example, warehouse employees often manage heavy loads, causing serious joint problems; pickers walk the length of almost a marathon each day and workers in cold chain warehouses frequently transition between normal temperatures and -30ÅãC. The unmanned warehouse technology has effectively improved working conditions.

JD.com began providing integrated technology services including both software and automation solutions to third parties in 2018. This opening up has demonstrated extraordinary results across industries such as FMCG, apparel, home appliances, home furniture, consumer electronics, automotive, and fresh produce. In 2020, JD.com’s logistics business served more than 190,000 corporate customers.

JD.com’s unmanned warehouse technology has received intense media coverage since its deployment. It was featured on the April 21, 2018, broadcast of China Central Television’s nightly news program Xinwen Lianbo. This program is shown simultaneously by all local TV stations in mainland China, making it one of the world’s most-watched programs. The warehouse has also been covered by leading international media outlets including CNBC, Bloomberg, The Washington Post, and others. 

Furthermore, as a representative of contemporary Chinese science and technology, JD.com was also featured during the closing ceremony of the 2018 PyeongChang Winter Olympics in an 8-minute promotional video for the 2022 Winter Olympics, which will be held in Beijing.