Chinese retailer JD.com says it has the world's first fully automated e-commerce warehouse. The Shanghai warehouse uses robots developed by Mujin, a spin-out from Japan's Tokyo University.
For some time now, several companies have been deploying robots on a large scale to automate operations in logistics centres and warehouses. One example is e-commerce giant Amazon, which even incorporated Automated Guided Vehicle manufacturer Kiva Systems in 2012. The robots are used by Amazon to transport entire racks of products to human workers, who remove the required products from the racks and pack them.
Warehouse of JD.com continues
JD.com's warehouse goes further and is fully automated. Robots are used to retrieve products from the warehouse, package these products and load the packaged products onto a truck, with no humans involved. CNBC reports that a total of 20 industrial robots are deployed in the facility, which is 40,000 square metres in size. The facility has been in operation since June.
These robots are controlled using a so-called robot controller, which acts as the 'brain' of the robots. Mujin argues that the realisation of such a robot controller previously stood in the way of fully automating warehouses. For instance, developing such an application is complex and therefore time-consuming. The company has therefore developed 'TeachWorker'. This software greatly simplifies the configuration of robots. Among other things, TeachWorker does this by collecting the values of various robot links in clear lists, providing insight into the robots' operation via 3D views, and using a proprietary language called Industrial Task Language (ITL).
Motion planning technology
ITL is not so much designed to programme specific movements of a robotic arm, but rather to make it perform a particular task. TeachWorker uses 'motion planning technology' to calculate how best to perform this task. Users only need to specify a start and end point, after which TeachWorker independently calculates how to reach this end point. In the process, the robot's surroundings are closely monitored to detect obstacles such as other robots and avoid collisions.
In particular, the technology allows robots to learn simpler movements on an immediate basis and without learning time. Mujin argues that TeachWorker also allows robots to be deployed in tighter space than is possible with traditional robot controllers.
No knowledge or experience needed
In addition, TeachWorker can increase the efficiency with which robots perform work. Mujin argues that this efficiency has traditionally depended significantly on the programmer who teaches the robots operations. The more experienced this person is, the more efficiently robots tend to move around. TeachWorker breaks this relationship and in its first 'run' calculates the best course of action to maximise the robot's efficiency. This requires no knowledge or experience.
An offline simulator can be used when programming the robots. This software gives users insight into the operation of their robots without actually activating them. The software also helps test changes in the layout of a warehouse - for example, by adding additional robots - in advance and understand their impact. The simulator provides support for CAD and CAM, allowing instructions for tasks such as welding, laser cutting and deburring to be generated by simply uploading CAD data to the controller.
Only five people employed
By the way, the fact that the warehouse is fully automated does not mean that there are no humans in the facility at all. CNBC reports that a team of five employees is employed to maintain the robots. All other work is carried out by robots. TeachWorker also allows remote maintenance to be carried out via the internet.
The project was announced by Mujin and JD.com last year. The parties then released the video below.
Author: Wouter Hoeffnagel
Source: Mujin
Source: CNBC
