The way we think of modern assembly lines is far away from how they looked at their inception. Now, instead of a line of different persons each repeating the same task over and over again, we have automated mechanical arms. However, the concept has not changed: each robot on the assembly line repeats the same task or task set on each component that appears in its range of action
From Factory Floors to Robotic Arms: The Evolution of Assembly Lines
Indeed, assembly lines were among the first application for robots and automation. Unimate, the first industrial robot, was developed during the 50s and was installed at a General Motor factory in New Jersey in 1961. This first successful use of a robot on an assembly line set the stage for a steep acceleration. By the 1980s, robotic assembly lines became widespread across industries like automotive, electronics, and packaging. Nowadays, industrial robotics is fully integrated into the manufacturing processes.
And yet, while the presence of robots on the assembly line is over the half century mark, the opposite procedure, dismantling, is far from being fully and easily automated. Indeed, the significant variability and structural complexity of products remains a major challenge on the way to full automation.
But what makes dismantling so hard to automate? Can we not just reverse engineer the assembling process?
Image: ©RTDS
Why Dismantling Is Much Harder to Automate Than Assembly
As it turns out, we cannot. A dismantling robot would have to solve a series of problems that an assembling robot does not encounter on its path. First of all, the dismantling procedure usually deals with end-of-life products. This means the material will be damaged, rather than in the pristine conditions handled while assembling. This potentially damaged material could lead to breaking of parts. Moreover, end-of-life products might have missing parts or present hazardous material which require careful and specific handling.
Another difficulty arising in the disassembling process is the lack of standardized inputs. The products present on a dismantling line could be all different, including variable shapes. The original product data could be incomplete or missing. In certain case, a mixed material composition can complicate the separation procedure.
Lastly, dismantling includes the possibility of joints breaking, as they might have been welded or glued together. It also needs sequential dependencies requiring intelligent pathfinding, and a need for physics-based simulation to handle tight fit while avoiding breakage. These show how disassembly is not simply reverse assembly, but poses an inverse process challenge.
The above leads to these processes being characterized by their decision-making complexity.
A path towards solving this complexity can even include human-robot collaboration strategies to enhance flexibility, adaptability, and safety in dismantling tasks. Most of these solutions are currently at about Technology Readiness Level (TRL) 5 to 7, meaning the are validated in controlled or semi-structured environments.
Advances in AI can also help tackle this complexity. An AI system for disassembly should be able to handle dynamic task allocation between humans and robots, real-time adaptation to discovered product conditions, and multi-objective optimization (time, safety, material recovery).
RecAL’s Mission: Smarter Recycling for European Aluminium
As part of its goal of improving the circulation and usage of secondary aluminium in Europe, RecAL is working on an automatic dismantling cell prototype for aircraft sections.
The aeronautical industry uses different alloys for different parts of an aircraft frame. In the usual end-of-life process (even in a semi-automated one), parts end up all mixed together. This leads to contamination of aluminium alloys, changing the chemical composition of the material and its properties, often resulting in poorer performance in key features such as electrical conductivity or stress-resistance. This change in composition results in a loss in recycling and market value, as the final recycled material does not retain the needed key properties nor is chemically well-defined.
Above: Aluminium recycling at Alumil’s facility in Greece (© RTDS)
The RecAL automated dismantling line would sort the different components directly at the point of disassembly, ensuring purity and traceability, enhancing the market-value of recycled aluminium and reducing reliance on virgin materials, with the associated cuts in extraction and energy costs. This retained information is then included in another key component of the RecAl project: the digital product passport.
To implement an automated disassembly line, researchers from are pushing the boundary of the current state-of-the-art AI technology, developing and improving tools for various purposes: element detection (rivets, structural elements such as clips, stringers, etc., the frame itself to be dismantled, etc.), force/torque control, advanced robot control strategies such as reinforcement learning, and more.
Besides these aspects, researchers are also ensuring the developed AI system allows for more flexibility, for example in detecting unforeseen situations and subsequently adapting the dismantling strategies.
This new AI for dismantling will not be restricted to aircraft use. Scientists are sure the developed technology will be adaptable to other use cases, by considering the necessary characteristics of the target components and the overall disassembly context. As different tasks might be required, the already existing AI or machine learning models might need to be integrated through new training over relevant data sets. This will ensure robust performance in the specific application environment.
Image: ©RTDS
Towards Circularity: Smarter Disassembly, Cleaner Materials
An automated disassembly line powered by adaptable AI, capable of adjusting to each situation, product, or single element, is a very ambitious goal. But RecAL is set to deliver. By enabling traceability and preserving alloy purity at the point of disassembly, RecAL not only increases the market value of recycled aluminium — it also reduces Europe’s dependence on imported raw materials. This is a step forward not just for recycling technology, but for a truly circular, economically viable, and environmentally sustainable approach to aluminium use.
Do not miss any updates on AI-development and real-world applications. Subscribe to our newsletter here.
Cliò E. Agrapidis, PhD (RTDS Group)
