INCOST - Intelligent Construction Site Task Scheduler

Zenodo (CERN European Organization for Nuclear Research)

Abstract

INCOST (Intelligent Construction Site Task Scheduler) addresses a coordination problem specific to robotised deconstruction: on a construction site, physical workspace is a shared, exclusive resource. This work models workspace occupancy as a first-class spatio-temporal constraint within a symbolic planning framework. A hierarchical task network decomposes a deconstruction goal into primitive actions and allocates them to heterogeneous actors. The resulting action network is then encoded as a Simple Temporal Problem whose constraints include zone reservations derived from the allocation itself. The HTN fixes the decomposition and allocation, and the STP returns the time-optimal schedule for that fixed structure in polynomial time, trading global optimality for tractability and for schedules whose timing can be explained constraint by constraint. Zone reservations produced by one planning episode are fed back as blocked windows for subsequent planning, a schedule-derived map of dynamics (MoD) of site activity, so later operations are planned against the space the earlier ones will actually consume. Planning is implemented on the AIDDL framework (Köckemann, 2024, https://doi.org/10.5281/zenodo.14065311), which provides the modelling language, type definitions and solver integration used to express the HTN domain and the temporal network.This project is funded by the euROBIN Horizon Europe project (Grant Agreement No. 101070596) through its 1st Open Call subgrant INCOST (Agreement No. euROBIN_1OC_CP_9).

Authors 3

1. Pranav Shevkar Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University

2. Görkem Can Ertemli Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University

3. Sigrid Brell‐Çokcan Aachen

   RWTH Aachen University · Association for Robots in Architecture

   Affiliation as printed

   RWTH Aachen University

   Association for Robots in Architecture