AI-Driven Metro Induction Planning and Scheduling

Human heuristic operations within metro systems are nearing a tipping point both in the scale of the metro network and the complexity of the urban area. The Nagpur Metro case study shows that automated rostering with CP-SAT optimization and telemetry based risk assessment can produce dependable, auditable results in simulation without compromising the necessary safety headroom. The cases dealt with were strict safety based over-rided reasons (certificates, overhauls, Class-A defects) and there were no incidents on other criteria allowed for dynamic replanning within comparable limits within less than 30 ms of solver time in all the stored plans. This was achieved by an audit trail plus the written explanation and "what-if" analysis which led to the officer involvement being adequate without too many complicating factors: the officers assess the proposed plan, vote it through and approve for publication. An involved officer who understands the nuances of the case can question the suggestions made in a generic spreadsheet much better than a mere spreadsheet with figures. The future work on this topic could involve the adaptation to other networks and experimenting with the use of optical RFID readers to observe the consist information within the stabling bays. Another interesting avenue of future work would be energy-aware stabling to plan regeneration and battery thermal recovery for night storage thus allowing the metro concepts with batteries and no catenary to be shut down appropriately.

[1] The Metro Rail Guy, “Bids invited for Nagpur Metro Phase 2’s 48 coach contract RS-01,” Mar. 16, 2025. [Online]. Available: https://themetrorailguy.com/2025/03/16/bids-invited-for-nagpur-metro-phase-2s-48-coach-contract-rs-01/. Accessed: Oct. 19, 2026. [2] J.-F. Cordeau, F. Soumis, and J. Desrosiers, “Simultaneous assignment of locomotives and cars to passenger trains,” Oper. Res., vol. 49, no. 4, pp. 531–548, 2001. [3] G. Maróti and L. Kroon, “Maintenance routing for train units: The transition model,” Transp. Sci., vol. 39, no. 4, pp. 518–525, 2005. [4] P.-J. Fioole, L. Kroon, G. Maróti, and A. Schrijver, “A rolling stock circulation model for combining and splitting passenger train units,” Eur. J. Oper. Res., vol. 174, no. 2, pp. 1281–1297, 2006. [5] Q. Zhong, R. M. Lusby, J. Larsen, Y. Zhang, and Q. Peng, “Rolling stock scheduling with maintenance requirements at the Chinese high-speed railway,” Transp. Res. B, Methodol., vol. 126, pp. 24–44, 2019, doi: 10.1016/j.trb.2019.05.013. [6] D. Huisman, R. Freling, and A. P. M. Wagelmans, “Multiple-depot integrated vehicle and crew scheduling,” Transp. Sci., vol. 39, no. 4, pp. 491–502, 2005. [7] A. Caprara, L. Kroon, M. Monaci, M. Peeters, and P. Toth, “Passenger railway optimization,” in Handbooks in Operations Research and Management Science, vol. 14, C. Barnhart and G. Laporte, Eds. Amsterdam, The Netherlands: Elsevier, 2007, pp. 129–187, doi: 10.1016/S0927-0507(06)14003-7. [8] R. Freling, R. M. Lentink, L. G. Kroon, and D. Huisman, “Shunting of passenger train units in a railway station,” Transp. Sci., vol. 39, no. 2, pp. 261–272, 2005, doi: 10.1287/trsc.1030.0076. [9] J. T. Haahr, R. M. Lusby, and J. C. Wagenaar, “Optimization methods for the train unit shunting problem,” Eur. J. Oper. Res., vol. 262, no. 3, pp. 981–995, 2017. [10] N. Davari, B. Veloso, G. de A. Costa, P. M. Pereira, R. P. Ribeiro, and J. Gama, “A survey on data-driven predictive maintenance for the railway industry,” Sensors, vol. 21, no. 17, Art. no. 5739, 2021.