Technical Articles
Warehouse Robotics for SAP-Extended Warehouse Management, Part 3
The previous blog explained the technology underlying SAP Warehouse Robotics, the supported scenarios and outlined the major IT tasks. This blog outlines the steps you need to take to get it working. For a detailed description, your best source is the Github repository.
Bill-of-Materials: The things you need
To implement and run Warehouse Robotics for SAP-Extended Warehouse Management, you need the following pieces of software and materials:
- SAP Extended Warehouse Management (on-premise or hosted, only)
- Warehouse Robotics has been tested with SAP S/4HANA 1709 embedded EWM, SAP EWM 9.5 (expected compatibility with other EWM releases).
- Warehouse Robotics does not work with EWM in S/4HANA Cloud, for now.
- SAP EWM extension for SAP Warehouse Robotics (Open Source). This extension has 3 components:
- EWM-side OData Interface to robot-specific table extensions
- EWM Order Manager for deployment in Google Cloud Robotics
- EWM Robot Controller for deployment on Google Cloud Robotics
- Recommended, but not mandatory: SAP Cloud Connector and SAP Cloud Platform
- Recommended for reasons of IT governance and IT security
- Alternatively, the SAP Gateway of EWM system can be exposed for incoming connections from Google Robotics Cloud. This has not been tested, however.
- Access to Warehouse Robotics via Google Cloud Platform (GCP)
- A robot, typically with additional equipment such as carts or docking stations
- Initial focus is on warehouse automation. For a list of already supported OEMs please contact SAP
- Effort for non-supported OEMs is low, in particular for those compatible with ROS.
If this is your first robot, there are a number of additional tasks unrelated to EWM. A robot is a new machine and like with every new machine in an industrial environment you need to get familiar and define rules. Here are some of the things to do:
- It is helpful to have a “robot nanny”: An individual in the team who becomes familiar with the robot hardware and software provided by the robot vendor.
- Create a SLAM map with the robot and identify points of interest. Depending on the robot, QR-codes attached to walls or columns may be a good option.
- A corporate security officer may need to grant approval before the robot can operate.
- Forklift drivers, packers, and other individuals need instructions on working with the robot and how to avoid dangerous situations.
- Identify areas the robot must or should avoid such as busy aisles
How to implement
For implementation, we recommend SAP implementation services. The subsequent steps illustrate the implementation of two scenarios (see the chapter about supported scenarios in the previous blog) in a single warehouse. The chosen scenarios are cross-docking and put-away.
Roles are:
- LoB: The line of business operating the warehouse using EWM
- IT: The organization technically running the EWM system, performing system maintenance, upgrades, etc. In some organizations, such tasks are performed by IT consultants
- EWM expert: An individual with in-depth SAP-EWM knowledge
- Cloud Provider Support: Customer support by your Cloud Provider (such as Google Cloud Platform)
Preparation
# | Task | Task owner |
1 | Ensure that items from Bill-of-Materials are available | IT |
2 | Define the storage area(s) for Put-away scenario | LoB |
3 | Prepare storage area
|
IT |
4 | Define the scope of tests, test cases, and success criteria for the start of productive operations | LoB |
5 | Install SAP Cloud Connector, if needed | IT |
Implementation in Sandbox
# | Task | Task owner |
6 | Ensure technical access to Cloud Platform | IT |
7 | Import the EWM Extension for Warehouse Robotics into Sandbox | IT |
8 | Code adaptations to the Extension, when using old versions of EWM, only | EWM expert |
9 | Implement SAP Cloud Connector for the sandbox and test connectivity | IT |
10 | Test in the sandbox, in the Q-system | IT, LoB |
11 | Create end user documentation | LoB |
12 | Create training material for end users and schedule training | LoB |
13 | Perform training for key users and end users | LoB |
The core effort for implementation (tasks # 6 – 9) can be performed within a week.
Implementation in Productive System
# | Task | Task owner |
14 | Prepare Cloud Platform project for productive operations | IT |
15 | Set up SAP Cloud Connector for the productive system and test connectivity | IT |
16 | Release project | IT |
17 | Go-live transport from Q-system into the productive system | IT |
In the end, your robot is operational. While these steps still appear like some work, consider the effort for an additional robot: Assuming that access to the robot’s system is established, you would need to just share the SLAM map and follow the instructions to connect a robot to the cloud. Assuming the label of the new robot matches previously rolled out controller applications, those will be deployed locally on the robot.
thanks for sharing this information, very informative and detailled.
i have a question, so SAP Cloud Robotics still requires integration with Google Cloud Platform?
The solution can run on any cloud platform that can run Kubernetes, because all the code is Open Source.
Hi Albrecht,
Do you for the lab purpose which robot can use which has a low cost or can we integrate the LEGO Mindstorms or any local robot with this architecture. What is the minimum requirement of the robot? For me cost is the main priority as this is only for the inhouse lab.
Regards,
Anil
Hi Anil,
thanks for asking! The robot needs to run a Kubernetes Cluster.
For the MiR100 and the MiR250 (both by Mobile Industrial Robots), we installed an Intel NUC on the robot. We do not deploy the Kubernetes Cluster on the manufacturer's compute unit and therefore have no liability risks. Such liabilities could emerge if we would tamper with the original hardware. By using a separate compute unit we avoid this.
Of course, there are alternatives to the Intel NUC; such compute units are commodities.
By the way, stay tuned for a SaaS solution next year, which makes all of this even easier.
Kind regards,
Albrecht
Hello Dr. Albrecht,
Thanks for the reply. I have one followup question from your reply.
I read your old comment in the blog part 2 "To be more precise, edge devices do not necessarily need to connect to GCP but to the Cloud Kubernetes Cluster. The setup introduced here assumes a deployment in GCP, but it would be possible to deploy somewhere else."
Does it mean that GCP is optional and we can directly connect the EWM(with Add-on) to Robot which has "Kubernetes Cluster"?
Also, Can you please share some more detail about "SaaS solution"? Is it possible that SAP Partner might also involve early in this initative?
Best Regards,
Anil Kumar
Hello Anil,
You could also run a Kubernetes Cluster on-premise; However, you cannot connect the robot directly to EWM.
SaaS = Software-as-a-Service.
Kind regards,
Albrecht
Hello Dr. Albrecht,
One funny question :): Is it technically possible to use other hyperscaler like (AWS or Azure) instead of GCP ? If yes, then can you please provide us some insight.
Best Regards,
Anil