In-Station Train Dispatching: A PDDL+ Planning Approach

Abstract

In railway networks, stations are probably the most critical points for interconnecting trains' routes: in a restricted geographical area, a potentially large number of trains have to stop according to an official timetable, with the concrete risk of accumulating delays that can then have a knockout effect on the rest of the network. In this context, in-station train dispatching plays a central role in maximising the effective utilisation of available railway infrastructures and in mitigating the impact of incidents and delays. Unfortunately, in-station train dispatching is still largely handled manually by human operators in charge of a group of stations. In this paper we make a step towards supporting the operator with some automatic tool, by describing an approach for performing in-station dispatching by means of automated planning techniques. Given the mixed discrete-continuous nature of the problem, we employ PDDL+ for the specification of the problem, and the ENHSP planning engine enhanced by domain-specific solving techniques. Results on a range of scenarios, using real-data of a station of the North West of Italy, show the potential of our approach.

Taming Discretised PDDL+ through Multiple Discretisations

“…PDDL+ is the most expressive formalism of the PDDL family of languages, which also includes PDDL (McDermott et al 1998) and PDDL2.1 (Fox and Long 2003). The modelling capabilities of PDDL+ have enabled the use of automated planning to solve complex real-world problems such as traffic control (Vallati et al 2016;El Kouaiti et al 2024), safety requirements for cyber-physical systems (Aineto et al 2023), train dispatching (Cardellini et al 2021), unmanned aerial vehicle control (Kiam et al 2020), pharmacokinetic optimisation (Alaboud and Coles 2019), and popular video games (Piotrowski et al 2023). Finding solutions for PDDL+ problems remains a daunting challenge due to its expressive power, further compounded by the scarcity of planners capable of effectively handling them.…”

“…A more advanced approach, presented by (Ramirez et al 2017) and supported by ENHSP, is to consider two different deltas: a simulation delta, to be as small as possible to better approximate complex hybrid dynamics, and a planning delta, that can be discretionally large, to reduce the burden on the planning process by avoiding decision points when no actions are likely to be applicable. An approach similar in nature, but domain-specific, has been proposed for the Train Dispatching Problem (Cardellini et al 2021), where the ENHSP planner has been modified to skip irrelevant decision points when controlling the dispatching of trains.…”

Reformulation Techniques for Automated Planning: A Systematic Review

“…Within this discipline, domainindependent planning refers to those approaches that keep the knowledge model, the domain knowledge related to the problem at hand, separate from planning logic, that enables automated reasoning to generate plans. The development of domain-independent planners within the AI Planning community facilitates the use of this "off-the-shelf" technology for a wide range of applications, including UAV manoeuvring [Ramírez et al, 2018], space exploration [Ai-Chang et al, 2004], and train dispatching [Cardellini et al, 2021]. This is despite the complexity issues inherent in plan generation, which are exacerbated by the separation of planner logic from domain knowledge.…”

Reformulation techniques for automated planning: a systematic review

“…Within this discipline, domain-independent planning refers to those approaches that keep the knowledge model, the domain knowledge related to the problem at hand, separate from planning logic, that enables automated reasoning to generate plans. The development of domain-independent planners within the AI Planning community facilitates the use of this 'off-the-shelf' technology for a wide range of applications, including UAV maneuvering (Ramírez et al, 2018), space exploration (Ai-Chang et al, 2004), and train dispatching (Cardellini et al, 2021). This is despite the complexity issues inherent in plan generation, which are exacerbated by the separation of planner logic from domain knowledge.…”

“…, 2004), and train dispatching (Cardellini et al. , 2021). This is despite the complexity issues inherent in plan generation, which are exacerbated by the separation of planner logic from domain knowledge.…”