Development history
September 20th, 2019 to October 3rd, 2019 Project got funded by the US DOE. Prepared to move from A-RESCUE to EV simulation.
October 11th, 2019 to November 11th, 2019 Implemented the charging station and electric vehicle classes.
November 11th, 2019 to December 11th, 2019 Added bus class and bus charging behaviors.
December 12th, 2019 to February 11th, 2020 Conducted energy model profiling.
February 11th, 2020 to March11th, 2020 Addressed ConcurrentModificationException.
March 11, 2020 to June 2, 2020 Developed the HPC framework and implemented the multi-arm bandit online routing algorithm.
June 3, 2020 to September, 2020 Implemented eco-routing for electric buses.
September 3, 2020 to January 2021 Transferred HPC module from JAVA to Python.
January 2021 to October 2021 Transferred online opertional algorithms from Java to Python.
October 2021 to February 2022 Moved the project to Github. DOE project ended.
February 2022 to May 2023 General bug fixing and code cleaning, added traffic light/signal supports
May 2023 to January 2024 Project got funded by the US DOT. Created SUMO network loader.
January 2024 to March 2024 Introduced Kafka for a more realistic data stream modeling.
March 2024 to April 2024 Added the support of simulating private vehicles’ trip chains.
April, 2024 to June 2024 Reimplemented the synchronized mode by adding interactive APIs and co-simulation runner with CARLA.
July 2024–February 2025 Built the Docker image for METS-R SIM, enabling its use within the METS-R HPC program.
February 2025–March 2025 Majorly refactored the METS-R HPC module. Switched from a custom launcher to Docker-based execution for more scalable and portable simulation management. Debugged all existing interactive APIs and released a comprehensive
interactive_example.ipynbtutorial notebook covering step, query, and control workflows. Added taxi dispatch and transit control APIs in the Python client.March 2025 In METS-R SIM: removed legacy eco-routing and dynamic bus scheduling code (moved to the HPC layer). Changed trip generation behavior so that requests for unavailable vehicles are queued rather than silently dropped.
April 2025 Replaced the NYC shapefile-based road network with a higher-accuracy SUMO
.net.xmlfile, adding smooth transition curves within intersections and refined lane-changing behavior. Added a pre-charging heuristic for private EVs before long-distance trips. Expanded CARLA co-simulation map support to all 10 CARLA Towns. Added charger type preselection (L2/DCFC). Updated EV utility functions based on the latest EV survey data. Added security attack example notebooks to METS-R HPC. Improved the efficiency of the neighboring graph construction.May 2025–June 2025 Overhauled co-simulation vehicle handling: added
exitCoSimRegionAPI, updated vehicle waypoint reporting inquery_coSimVehicle, and revised teleportation behavior. Added vehicle bearing updates when entering new lanes. Refactored private EV charging behavior and reorganized charging preference control logic. Improved runtime speed for vehicle relocation.July 2025 Major revision of the electric bus module: added
add_bus_route,add_bus_run,insert_bus_stop, andremove_bus_stopcontrol APIs for dynamic bus route management; updatedquery_bus_routeandquery_route_busquery APIs. Added exception handling for routing failures on dead-end roads. Improved simulation performance by disabling the mode-split model as default.September 2025 Added dynamic EV charging price support. EV agents now prioritize low-price chargers during charging decisions. New
update_charging_pricescontrol API enables real-time price manipulation. Added a charging price attack scenario to the HPC security examples.October-December 2025 Implemented comprehensive traffic signal control APIs:
update_signal,update_signal_timing,set_signal_phase_plan, andset_signal_phase_plan_ticks. Extendedquery_roadto include downstream road information.February 2026 Added
query_centerlineAPI for lane-level geometry retrieval andquery_signal_groupAPI for full phase plan queries. Fixed traffic signal timing inconsistencies. Improved co-simulation vehicle entry handling and revisedquery_vehicleto correctly report vehicles located within intersections. Ensured full simulation reproducibility across all new features.March 2026 Added
save/loadAPIs for complete simulation checkpointing and deterministic replay. Resolved a data corruption issue caused by lane changing on curved roads. Fixed a race condition in the rerouting logic. Fixed duplicated vehicle IDs after callingreset().