In brief

A genetic algorithm helps automate and optimise path planning in laser-aided additive manufacturing, improving product quality, process efficiency and overall costs.

© Shutterstock

Ironing out kinks for laser-sharp precision

3 Jan 2023

An algorithm inspired by nature guides metal-depositing lasers for faster, more precise component manufacturing.

The forging of metals into weapons, jewellery and utensils is one of the oldest trades in human civilisation. Archaeological evidence suggests that skilled blacksmiths have been heating metals and hammering them into their desired shapes for thousands of years.

Today, lasers have replaced hammers. With laser-aided additive manufacturing (LAAM), layers of metal powders are melted and deposited by a high-energy laser beam, fabricating intricate shapes such as medical implants or aerospace components with unprecedented speed and precision.

Compared to traditional top-down manufacturing approaches, lasers in the LAAM process deposit metals from the bottom up by following manually guided routes called toolpaths. These toolpaths can range from simple parallel lines to more complicated contours that fill each layer in puzzle-like segments. However, common infill toolpath strategies often don’t fully account for geometrical kinks that may happen in the fabrication process.

“With complicated geometrical shapes, current LAAM approaches tend to form an undesirable bead and track features such as voids—awkward regions too narrow to fit another path in, leaving an irregular gap,” explained Youxiang Chew, a Research Scientist at the Singapore Institute of Manufacturing Technology (SIMTech). “These flaws give rise to defects that can degrade the mechanical performance of the final product.”

For engineers, planning and testing more efficient paths can be time-consuming and tedious. In response, Chew collaborated with researchers, Ning Liu from the Advanced Remanufacturing and Technology Centre (ARTC) and Yunfeng Zhang from the National University of Singapore (NUS) to develop an advanced algorithm that would help automate the planning process.

The research team’s new technique builds on previous path-planning approaches by using artificial intelligence (AI) to streamline the segmentation process. A genetic evolutional algorithm—a framework inspired by Darwin's Theory of Evolution—automatically divides the blueprint for a target object into multiple optimised sections. This approach allows the laser to deposit metal in the most efficient manner possible while minimising void formation.

“We based our approach on a genetic algorithm as it is more capable of ‘looking at the big picture’ to search for high-quality solutions to a complex problem like path-planning in LAAM,” said Liu.

The team found that the new algorithm was 90 percent faster than the current best approaches and produced consistently high-quality metal parts at higher deposition efficiency rates. These could translate to a significant reduction in the operating and labour costs of LAAM, making the technology more viable in other applications such as repair, remanufacturing and 3D printing.

The researchers are currently exploring how other AI techniques such as deep learning could further optimise the segmentation process. “These ‘learn-to-optimise’ techniques could provide near real-time optimisation results to improve the evolutional algorithm currently used,” concluded Chew.

The A*STAR-affiliated researchers contributing to this research are from the Advanced Remanufacturing and Technology Centre (ARTC) and the Singapore Institute of Manufacturing Technology (SIMTech).

Want to stay up to date with breakthroughs from A*STAR? Follow us on Twitter and LinkedIn!


Liu, N., Ren, K., Zhang, W., Zhang, Y.F., Chew, Y.X., et al. An evolutional algorithm for automatic 2D layer segmentation in laser-aided additive manufacturing. Additive Manufacturing 47 , 102342 (2021). │article

About the Researchers

Youxiang Chew has over nine years of experience in Laser Directed Energy Deposition (LDED) technology, focusing on processes, residual stress modelling and the integration of machine learning to optimise laser deposition toolpaths. Chew’s other research interests include developing high entropy alloys and bulk spatially patterned heterostructured multimaterials parts using LDED. Beyond the LDED process, Chew is now looking into new process innovation by combining different advanced manufacturing processes for multimaterial processing. The group is also exploring hybrid DED processes and new friction stir processing for repair and remanufacturing, aiming to streamline the manufacturing of high value components.
Ning Liu received his BEng degree in Mechanical Engineering and Automation from Xi’an Jiao Tong University in 2012 before being awarded a PhD in Mechanical Engineering from the National University of Singapore (NUS) in 2016. He is currently a Development Scientist at the Advanced Remanufacturing and Technology Centre (ARTC), Agency for Science, Technology and Research (A*STAR). Before joining A*STAR, Liu worked as a Research Fellow at NUS’s Department of Mechanical Engineering. Liu’s research interests include data-driven optimisation in supply chain operations, additive manufacturing solutions, industrial automation and sustainable manufacturing.

This article was made for A*STAR Research by Wildtype Media Group