[Use Case TF] Semi-Automated Warehouse

[samubura]

Title: Semi-Automated Warehouse

Submitter(s):

Samuele Burattini (University of Bologna)

Description:

A manufacturing facility has its last production step aimed at packaging batches of products in a package that can be stored. This then needs to be picked by either autonomous robots or employees and distributed in different locations of the storage facility so that they can be shipped at a later stage.

Human Employees and Robots have different load capacities (i.e. they can carry different amounts of weight) and can either be busy or idle. Whenever a new batch is ready the system will look for an idling picker and find means to instruct it to pick up the batch, this can be different of course as humans and robots may need to receive instructions using different means.

Employees are busy with different sets of tasks (e.g. maintenance). They have exclusive access to a special vault where only some products that need to be handled with care must be put.

An ML prediction model, trained on the measured times of different activities can offer estimates of the duration of the tasks and as such allow other systems to know when an employee will be free again. Similarly, robots have information on how long their route is expected to take until they are free again.

To optimize the system, when the production of a special batch is requested, it must be scheduled estimating when at least one employee will be ready, taking into account how long a batch takes to be produced. Moreso, the system should try to keep Robots and Employees as busy as possible and load them at full capacity.

Expected Participating Entities:

• Employees
• Robots
• Managing Agent
• Packaging Machine

Illustrative scenario(s):

1. The production of 10 batches is requested
2. The production starts and starts measuring the time for each batch to update its estimated time for batch
3. Whenever a batch is being produced, the system looks for a picker, prioritizing robots and trying to maximise load capacity
4. A special batch is requested
5. The system checks the estimated time per batch from the workshop and then surveys all employees. 5.1 if one is idling it will schedule the batch as the next one and contact that employee 5.2 if all are busy it will look for the forecasting service, find the one that will be free sooner and schedule the batch accordingly

The importance of Web Agents for the use case

Web Agents can enhance the flexibility required in this scenario, new employees or new robots can be added at any time, and they might have:

• different communication requirements
• different load capacities

To maximise the optimization of resources agents need to discover information at runtime and use it to accomplish their goals.

Existing solutions:

Existing solutions may require all robots and employees to follow the same interface and communication patterns and then hardcode the agents to use them. However, this is not always doable in practice and limits dynamic reconfiguration that might be beneficial for such kinds of systems.

Other information (optional)

This is a draft, just to start adding something to the use-case task force.

[Antoine-Zimmermann]

@samubura Thank you for the input for the use case task force! I have a question though: can you make it more precise why we need Web Agents specifically and not good old-fashioned Multi-Agent Systems?

[danaivach]

@samubura You mention that one of the advantages of using Web agents in this UC is to address different communication requirements, and then that existing solutions require all robots and employees to follow the same interface and communication patterns and then hardcode the agents to use them. Given the above, how would you position this UC with respect to the "Cross-Protocol Interaction in Industry 4.0 Scenarios" Use Case of the Web of Things (WoT): Use Cases and Requirements Group Note?

@Antoine-Zimmermann, as mentioned in the meeting, it might be useful that we all take a look at these Web of Things use cases, in general. They could provide some ideas for UC templates, but also help us to ensure that our UCs have some valuable delta from the WoT motivations and state of the art (e.g., we could take some of these use cases, extend them and see what are the benefits of considering the existence of Web Agents in these.) I see that @egekorkan has been a contributor of a WoT use case, so maybe he could comment on these use cases' relevance to the UC TF, or knows more about the motivation and the methods for coming up with this group note.

[egekorkan]

I see that @egekorkan has been a contributor of a WoT use case, so maybe he could comment on these use cases' relevance to the UC TF, or knows more about the motivation and the methods for coming up with this group note.

Today, we wrote an overall categorization with @Antoine-Zimmermann @andreiciortea @remcollier . We will share with the whole group in the next call :)

[danaivach]

Great, thank you all.