As businesses wrestle with the impact of artificial intelligence (AI) on their industries, Xavier Malina, Head of Artificial Intelligence at Ai Build provides his thoughts on the benefits AI could hold for advanced manufacturing processes like 3D printing.
At Ai Build, we and our customers spend a lot of time 3D printing large, complex, parts in polymers, metals and other materials. This leads us to brush up against the limitations of additive manufacturing, in terms of performance, reliability, ease of use and troubleshooting. Our experience and research have shown that the best way for the industry to overcome these limitations is to deeply integrate AI into the fabric of the manufacturing process at the software and hardware level.
Performance
In order for a part to print successfully – one must take all of the following into account: material, printing system capabilities (horizontal / non planar / multi axis), printing environment, and separately, the geometry, its intended function and the printing orientation. Today, this is done manually by 3D printing engineers. In the future, evolutionary AI systems will combine data from past prints with recordings from sensors, to simulate all possible toolpaths and pick the perfect one, tailored to a specific system in a specific environment. No more ‘one-size-fits-all’ slicing.
Reliability
Once you have the perfect toolpath, how do you make sure you can print five of these per day, every business day for the next quarter? The answer is an edge-AI print supervision system. These embedded systems will not only detect defects, but also predict defects before they occur and take live action to correct the defects before they go out of tolerance. For some materials, the system will be able to automatically evaluate errors and create a ‘repair toolpath’ for a tool like a CNC to correct the defects. No more waste.
Ease of use
It is a universal truth that industrial software packages for computer aided design and manufacturing have a steep learning curve. While beginners can use them quickly enough, experts alone can make the most out of them. Only experts know how to find those ‘secret’ menus and remember advanced approaches. Only experts can navigate towards a solution efficiently. No longer. With Large Language Model (LLM) technology, expert solutions become available to beginners – from their first interaction with the software. With tools like our Talk to AiSync, users will be able to type their intent in normal sentences – in any language – and the system will create a complete workflow appropriately, based on how experts would solve the problem. It will pay attention to the machine you are using, the material, ask follow-up questions, and make suggestions if it notices something suboptimal. Experts can modify settings directly, but beginners can shortcut learning by behaving like an expert immediately.
Get your FREE print subscription to TCT Magazine.
Exhibit at the UK's definitive and most influential 3D printing and additive manufacturing event, TCT 3Sixty.
Explainability, Trust, Troubleshooting
Why is this orientation the best way to print this part? Why is my part warping? What material should I use for this part? What caused that collision? Today, in-house experts and external-consultants are the only way to find answers on the factory floor. This can slow down the manufacturing process and lead to many repeat prints as you try and solve quality issues. In the future, LLMs like GPT4 will be fully integrated with slicers. Users will be able to receive answers that until now only an expert with deep knowledge of your tools could provide. This will help companies get set-up quicker. Even more important, LLM-powered AI systems will restore trust in the manufacturing process by enabling technicians to ask the system directly why something went wrong. No more black-box: LLM technology combined with Retrieval Augmented Generation architectures will finally deobfuscate the 3D printing process.
In summary, integrating AI into 3D printing processes brings several benefits. AI promises tailored toolpaths by eliminating the 'one-size-fits-all' approach. Preventive corrections help in reducing waste and saving time. There will be new intuitive ways to interact with complex interfaces and settings. Moreover, AI will provide increased transparency and trust by dispelling the complexity of 'black-box' manufacturing processes.
