Guangzhou Yunge Tianhong Electronic Technology Co., Ltd , http://www.e-cigaretteyfactory.com
Big data transforms traditional logistics and sees how to drive the “Second Spring Festivalâ€
In terms of the signing time of 100 million parcels, 2013 is 9 days; in 2014, 6 days; in 2015, 4 days, in 2016, only 3.5 days. This is the huge subversion brought about by “changing the track†– if you don’t need big data to transform traditional logistics, how do the major e-commerce platforms drive the double 11 logistics, which is called the “Second Spring Festival†campaign?
How to make pickers walk less <br> <br> double 11, vice president and division president of Jingdong Group X Xiao Jun warehouses have to experience the "ultimate peak" year.
"On this day, I walked up to 20 kilometers in the warehouse, and the picker went 50 kilometers in the warehouse on this day, which is equivalent to running a marathon." On November 23, Xiao Jun said at the Jingdong logistics brand strategy conference. .
Just like Facebook first met the needs of a group of boys who wanted to see photos of girls at Harvard University. How to make warehouse pickers less walking or even walking, this is what Xiao Jun originally wanted to do most. In 2014, Jingdong launched a “path optimization†system, which allowed warehouse pickers to walk an average of 15 kilometers less per day. The system picks up 5 kilometers per day for each iteration.
For example, the picker takes the S-shaped path, analyzes the big path of the past path, and then optimizes the path according to the position of the goods to be sorted now, so that the picker can go where and where to go.
"Path optimization" is a typical case of big data application to warehouse management. Here, the algorithm is the core.
The principle of Amazon is "do not go back." The Amazon backend has an algorithm that randomly optimizes the path for each picker. Every time a piece of goods is sorted and scanned, the scanner automatically tells the picker where the next item to pick is, and the picker goes straight ahead without going back. This mode ensures that the pickers have the least walking after they have been selected, at least 60% less than the traditional mode.
This is due to Amazon's "random shelves" rule. When the product is put into storage, the goods are stored based on the size classification. Under the premise of reasonable classification of goods, goods with similar attributes will be placed on similar shelves. For the goods in the process of continuous picking, you can place where you have time, which greatly reduces the time for selecting the goods, and you can walk less.
Trouble will not be left to the late picking process? Still relying on big data. Due to the dispersed distribution of goods, it is possible to combine the most adjacent products and the different products ordered by a single customer into an optimal path from the location of the entire inventory.
Amazon is the fastest 30 minutes from order to delivery, picking, packaging, and delivery, but Chinese peers have already set a new record because they are facing a much larger Chinese market than Amazon. They need to be faster. The speed allows buyers to remember their quality service. Here, Jingdong’s record is 2 minutes.
This year's double 11, the fastest one in Jingdong, it took only 12 minutes from Shanghai buyers to order delivery. This is not a show, big data "pre-clothing" can be easily done.
The specific method is that the back-end system records the trajectory and history of the products that customers browse, and analyzes the customer's needs through big data, and then puts the products that are of interest to customers in the warehouse nearest to them.
“Advance delivery†has become the standard for logistics warehousing. SF has a similar product "Shunfeng Lighthouse". Not long ago, Jingdong established the Y Division to build a smart supply chain, in which the sales forecasting platform is specially designed for early delivery.
Amazon played even more, and in 2014 it received a pre-planned ordering patent, which was shipped before the user placed the order based on the user's buying habits and hobbies. The system can ship the package to the destination area without specifying a specific shipping address and time, and the specific address will be determined during transportation. If you make the right decisions, you can reduce logistics costs by 10% to 40%.
“Which goods can be put together to improve the sorting rate and which goods are stored for a short period of time can be rationally arranged through the analysis of the correlation between commodities through big data.†Xiao Jun said that now, the number and area of ​​warehouses are With double growth, reasonable placement of goods is of great significance in improving warehouse utilization and handling sorting efficiency, but in the past it was only based on experience.
The training takes AlphaGo sorting robot as smart as <br> <br> data show that in the transportation and warehousing logistics system composed of warehousing logistics costs account for 60% of total operating costs, improve inventory turns, reduce logistics company The key to cost.
The "Jingdong No. 1" intelligent sorting center, which was put into operation this year, can process millions of orders per day, and the sorting efficiency is increased by more than 5 times, and the labor cost is saved by more than 70%. The unmanned warehouse developed by JD.com has applied various automation technologies and even magnetic levitation technologies in the industry, and has also built a system solution to support data sensing and artificial intelligence algorithms on the devices. Excellent learning ability.
The unmanned warehouse is equipped with four explosive robots, which are DELTA sorting robots with 3D vision system; intelligent handling robot AGV with inertial navigation and automatic obstacle avoidance; SHUTTLE shelf shuttle with high running speed and accurate positioning; The six-axis robot 6-AXIS with a maximum of 165 kg and a wingspan of nearly 3 meters.
“What kind of robot is used to help production and help which part to improve efficiency? This comes from Jingdong’s accumulation of understanding of logistics and warehousing industry for more than 10 years.†Xiao Jun said that in the process, many partners are needed to help us innovate together. .
Xiao Jun's words are not difficult to understand. In the eruption period of the e-commerce market, a trillion-level logistics and warehousing industry has been erected. The thinking of partners and the open situation have become a consensus.
In reality, it takes 36 seconds for a package to be sorted by a person, and only 1 second for a robot operation, and the efficiency is increased by 35 times. The key point is, “Traditional sorting centers should have more people staring at the parcels. This is a great challenge to the physiological limit when the double 11 is promoted.†Xiao Jun said that the sorting robot is under the constant training of people. Deep learning gains this powerful ability.
This kind of training is just like Google training AlphaGo. Google uses hundreds of thousands of Go game masters to train and make it smart. In computing, it uses tens of thousands of servers to train AlphaGo to play chess. And let the different versions of AlphaGo play tens of millions of discs with each other, which guarantees that it has no idea. Compared to this, training the sorting robot is a piece of cake. Jingdong's self-operated goods and SKUs have more than 30 million. Jingdong installed a camera in the manual storage center, allowing the robot to learn how to bind the goods according to the operator's operation mode every day. After a week of training, the robot recognizes a product after 100 cycles, and the picking accuracy rate will reach 99% from 80% in the past.
"We will also train robots to work in a variety of lighting environments, sometimes to supplement the light, sometimes dim light, and then optimize, so that the robot can be applied to a wide range of work." Xiao Jun said, this is also human understanding robot One way. 
In 2016, the double 11, 157 million parcels were linked together, equivalent to 380,000 kilometers from the earth to the moon, and the distribution distance can add up to the solar system. In the next few years, the number of Chinese express delivery orders will reach 200 million. This means that even if the traditional logistics puts more manpower, it will not be able to keep up with the non-linear growth rate of the e-commerce outbreak.