Machine learning (ML) offers a powerful solution to replicate and optimize complex physical systems, presenting an extraordinary opportunity for revolutionizing packaging design.
Amcor is seeking a partner to create a model to automate and optimize the PET (polyethylene terephthalate) container design process, setting new industry standards in the development and manufacturing of custom PET containers.
In 2-step stretch blowmolding of PET (polyethylene terephthalate) containers, the first step is to injection mold the plastic into the shape of a preform, which looks like a test tube with a threaded neck. The design of the preform enables the precise distribution and molecular orientation achieved in the next step, where the preform is then heated and placed in a blow mold. High-pressure air is injected into the preform, causing it to expand and take the shape of the surrounding mold, resulting in the final PET container.
Amcor has a database of successful preform designs and their corresponding container outcomes, but this valuable data has not been utilized to streamline or enhance the design process automatically. Given Amcor's extensive database and experience in this field, there is a unique opportunity to harness this data to develop a dynamic and continually improving AI model.
What we're looking for
We are looking for an AI model that can automate the design process with high level of confidence. Additionally, the model should be modifiable so we can incorporate new data and attributes of interest as they become available.
Solutions of interest include:
Artificial Intelligence
Machine Learning
Automation
Our must-have requirements are:
Examples of previous models demonstrating effectiveness of the proposed solution
Must be modifiable to incorporate future data
Our nice-to-have's are:
Solutions that can integrate with existing design software (CAD systems), allowing the translation of AI design outputs into practical CAD drawings, enabling direct manipulation and visualization of designs
We are able to provide funding for model development. Final funding amounts to be discussed at the appropriate stage of engagement.
Expertise
Partners will have access to experts from the industry (CAD engineer, product development engineer, polymer scientist etc.) to guide them to best practices and current manufacturing processes.
Tools and Technologies
Partners will be able to access our work processes and training materials to understand the development cycle and manufacturing processes.
Data
Selected partners will be able to access to our CAD data as well as real life test data for the duration of project.
Facilities and Services
Partners will be able to access to our pilot plant and analytical labs to generate relevant data.
Who we are
We solve packaging challenges, around the world every day. We develop differentiated products, services and processes to protect your products and the people who rely on them, all around the globe. Drawing on unrivaled heritage in design, science and manufacturing, more than 1000 R&D experts are innovating new materials, formats and technologies to better protect your products.
In 2018 we pledged to develop all our packaging to be recyclable or reusable by 2025. On our journey to our 2025 pledge we are innovating across various sustainability options, delivering you more sustainable packaging solutions under the new EcoGuard™ brand.
Hello, if the research leads to publishable results, is it possible to publish them while anonymising the source? If not, what would be the benefit to us as a university of collaborating? Thanks
Amcor is open to sponsoring research for a PhD candidate or a post-doctoral researcher. As part of the JDA agreement, we will establish which party owns the technology as related to the processing and as related to the software. The agreement will be clear prior to the funding and the start of research. Thanks!
At the moment, we do not have an NDA in place and I am not able to provide data. If you are able to reference other similar work and successes or your logic plan for working with the data, that would be excellent.
Just to understand the project scope and interface, could you comment on the expected specific output – would this be like a Chatbot? Would the aim be that a user enter an input, and the system will generate the design?
Hello Carolyn and thank you for your question.
The approach is open for discussion. Container blow molding is dependent on design, but it is also dependent on material properties, stretching rates, heating, etc. Data is generated and collected which is currently evaluated manually. Establishing the final design and the final blow processes is an art.
We are hoping to use the data we currently collect to (1) generate new designs and (2) identify the optimal operating space
Hi Gregory,
This is a great topic, thanks for the outstanding formulation. We have two questions:
The data you have is it only experimental based or would your simulation (FEA) team be able also to provide physics data?
Second question: in the application process we would like provide case studies that we have done with previous customer at Digimindlabs.com. How can we add the case studies to the application?
Thanks
Omar
Hello,
There should be a way to cite publications.
The Halo proposal is meant to be a brief summary, but we can have a conference call to discuss your presentation. Please mention the presentation in the proposal so that I can schedule a call with our team.
Thanks!
Hello, does this project aim to optimize the operation parameters or container design, or both? I just submitted a brief proposal and included the joint publications with industry. If you would like to hear more details of these projects, I will be happy to do a brief presentation online.
Hello, thank you for submitting the proposal. We will review as a team and let you know if we have any questions. If so, we can schedule a call to discuss. Thanks again!
We are continually creating new container designs which enhance visual appeal, enable light weighting, and improve product-package interactions. We are looking for a solution that will continually optimize both the design of the preform and the blow molding process as new data and as new processing technologies becomes available. Does this answer your question?
Not really.
I'm asking: 1) when would you expect research projects to start?
2) when would you like to see them conclude? (academic projects are typically multi-year projects, but I also represent a software solutions company that can address projects in less time).
Following review of your proposal, we will need to establish an NDA / JDA. I would hope to start by September and to complete the initial round within 1 year. Hopefully, we will begin to identify additional areas to explore as we develop together.
Looking forward to seeing your proposal.
Hi Gregory,
I represent an ML-oriented data science team based in Europe, Slovenia, multiple PhDs, that's interested in scheduling a call to discuss tackling the project. Would you be available for such a call?