The Travelling Salesman Problem (TSP) is a classic combinatorial optimization problem with many real-world applications. For large datasets (over 30,000 nodes), TSP is usually tackled using heuristic or nature-inspired metaheuristic algorithms.

Recent research on Ant Colony Optimization (ACO) has shown promising results on large-scale TSP instances.

This project builds on that work by:

Reviewing and analyzing existing properties and improvements of ACO.

Proposing two new enhancements:

Controlling ant behavior by forcing path changes.

Dynamically adjusting the number of mobile ants over time, combined with smooth Max–Min updating.

Achieves less than 0.3% error on Art TSP datasets with 100,000–200,000 nodes.

Achieves less than 0.8% error on other datasets with 33,810–100,000 nodes.

Outperforms the current state-of-the-art ACO algorithms.

Discovered new best-known solutions for several benchmark datasets.