LG AI Research Studies PCB Automatic Routing Technology

PCB (Printed Circuit Board) design, an essential component of all electronic products, centers on the routing stage, where circuits are connected after parts placement. However, the routing process involves highly complex constraint conditions, such as wire thickness according to current amount, spacing to prevent short circuits, and multi-layer design.

To solve this problem, LG AI Research uses two complementary methods.

• Grid-based method: Explores by dividing the board into regular intervals, converting the problem into a Steiner Tree Packing Problem (STPP) to quickly find paths that satisfy the majority of constraint conditions.
• Continuous Space-based method: To resolve errors that occur during grid conversion, this method uses the actual board as-is and performs precise routing by modifying the RRT (Rapid-exploring Random Tree) algorithm.

In particular, by combining the Pathfinder+ algorithm—which uses grid-based search results as a guide to narrow down the search area—with RRT-PCB, both search efficiency and accuracy were improved simultaneously.

This technology has currently been validated through actual products from LG affiliates, and there are plans to further advance it by learning the know-how of experienced engineers, aiming for a level that can be immediately applied in actual design settings.