What the study found
The study found that a tightly integrated mix of exact optimization and heuristic search produced state-of-the-art results for the bus driver scheduling problem, which is the task of designing driver shifts to cover planned bus tours under legal and contract rules. The authors report exact solutions for small instances and low gaps to a known lower bound for mid-sized instances.
Why the authors say this matters
The authors say the approach provides high-quality solutions across instances of different sizes, and they conclude that the tighter integration of Large Neighborhood Search, a method that makes partial changes to a solution, with Column Generation, a method that builds solutions from promising pieces, can improve on a version where Column Generation is used only as a black box. They also state that the methods are general and may apply to other rule sets and related optimization problems.
What the researchers tested
The researchers studied an exact Branch and Price method and a Large Neighborhood Search framework that uses Branch and Price or Column Generation in the repair phase. They also proposed a deeper integration of Branch and Price and Large Neighborhood Search by storing columns generated from subproblems and reusing them in other subproblems or to search for better global solutions.
What worked and what didn't
The evaluation showed that the approach delivered new state-of-the-art results for instances of all sizes. According to the abstract, Branch and Price performed best on small instances, while the tightly integrated Large Neighborhood Search and Column Generation approach produced high-quality solutions for larger instances and improved over Large Neighborhood Search that used Column Generation only as a black box.
What to keep in mind
The abstract does not describe detailed numerical results, runtime values, or a full list of limitations. It also does not explain which specific rule sets beyond complex break constraints were tested, although it says the methods are general enough to be applied more broadly.
Key points
- The paper reports state-of-the-art results for bus driver scheduling with complex break constraints.
- Exact solutions were obtained for small instances, and low gaps to a known lower bound were reported for mid-sized instances.
- Branch and Price performed best on small instances, according to the abstract.
- A tighter integration of Large Neighborhood Search and Column Generation improved results for larger instances.
- The authors say the methods may also apply to other rule sets and related optimization problems.
Disclosure
- Research title:
- Hybrid scheduling methods improve bus driver assignment results
- Authors:
- Lucas Kletzander, Tommaso Mannelli Mazzoli, Nysret Musliu, Pascal Van Hentenryck
- Institutions:
- TU Wien, Georgia Institute of Technology
- Publication date:
- 2026-04-20
- OpenAlex record:
- View
- Image credit:
- Photo by Bagzhan Sadvakassov on Unsplash · Unsplash License
Get the weekly research newsletter
Stay current with peer-reviewed research without reading academic papers — one filtered digest, every Friday.