Traffic & transit simulation
Build and calibrate traffic models to observed data.
20 project-authored proposals · taxonomy v1.1 · 0 new benchmark runs. Verifiers and partial-credit rules below are designs, not executed results.
- Typical applications
- SUMO · MATSim · PTV Vissim
- Class GUI-need label
CL· CLI-sufficient
Task proposals · 20
Expand to inspect the complete original wording and proposed checks. Original proposal text remains English in both interfaces.
R07.13.01.01-T01 Run a SUMO scenario and report travel-time statistics
Original instruction
Run SUMO 1.20 on fixtures/downtown/downtown.sumocfg (seed 42) with tripinfo output, then write output/stats.json with vehicles arrived, mean travel time (s), mean waiting time (s) and the 95th percentile travel time.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
Grader reruns with seed 42: counts exact, means within 0.5%, p95 within 1%. Config hash checked.
Proposed partial credit
25% arrivals; 50% means; 25% p95.
R07.13.01.01-T02 Calibrate route flows to detector counts with GEH under 5
Original instruction
For fixtures/arterial/ (SUMO net, 12 induction-loop counts in input/counts_0800.csv for 15-min intervals), generate routes with routeSampler or flowrouter so that simulated counts give GEH <5 for at least 85% of detector-intervals. Save output/routes.rou.xml.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Grader simulates 3 seeds with the routes: share of detector-intervals with GEH<5 >=85% in each seed; routes valid (all edges connected).
Proposed partial credit
20% valid routes; 80% GEH share (pro-rata from 0.50 to 0.85).
R07.13.01.01-T03 Calibrate an AM-peak network to counts and travel times
Original instruction
Calibrate fixtures/metro_am/ (SUMO) to input/counts.csv (40 detectors) and input/travel_times.csv (6 corridors) by adjusting OD scaling per district and car-following parameters (tau, sigma, minGap within input/bounds.json). Targets: GEH<5 on 85% of counts and corridor times within 15%.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO · Python
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
long· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
Grader reruns 5 hidden seeds: GEH share and corridor time errors averaged across seeds; parameters within bounds; network unchanged.
Proposed partial credit
10% bounds respected; 45% GEH target; 45% travel-time target (pro-rata).
R07.13.01.01-T04 Draw a new signalized intersection in netedit
Original instruction
In netedit, open fixtures/suburb.net.xml and build the new 4-leg intersection of Oak Ave and 5th St per input/intersection_plan.pdf: lane counts, dedicated left-turn lanes, turning connections and a 4-phase traffic light program (90 s cycle). Save the network.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO · netedit
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clickdrag· Drag & droppath· Continuous pathgeom· Precise geometry- Work product
WP04· 3D/spatial model- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Grader parses the net: junction type traffic_light, lane counts and connections match the plan (set equality), TLS program 4 green phases summing with yellows to 90 s; netconvert check passes. The agent VM has no terminal, so netconvert and plain-XML edits are unavailable; the file audit log must show the .net.xml written by netedit.
Proposed partial credit
30% geometry and lanes; 30% connections; 30% signal program; 10% clean network.
R07.13.01.01-T05 Retime an intersection with Webster's method and cut delay
Original instruction
For the isolated intersection in fixtures/webster/ (SUMO, fixed-time 120 s), compute a Webster-optimal cycle and green splits from the flows in input/flows.csv and saturation flow 1,800 veh/h/lane, implement it in the TLS program, and show average delay drops by at least 15%. Write output/timing.json.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO · Python
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Grader: cycle and splits within 2 s of Webster solution; reruns 3 seeds: mean delay reduction >=15% vs baseline.
Proposed partial credit
40% Webster timing correct; 20% implemented in net/additional; 40% delay reduction.
R07.13.01.01-T06 Build a roundabout model with conflict areas in Vissim
Original instruction
In PTV Vissim, build a single-lane roundabout (inscribed diameter 40 m) on the background image input/roundabout_aerial.png (scaled with the given reference line): links and connectors for 4 approaches, conflict areas with circulating priority, vehicle inputs and static routes from input/turning_counts.csv. Run 3,600 s and export max queue per approach.
- Platforms
win· Windows- Applications
- PTV Vissim
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickpath· Continuous pathdrag· Drag & dropgeom· Precise geometry- Work product
WP04· 3D/spatial model- Conditions
licensed· Licensed pro app- Proposed difficulty
- 4 / 5
Proposed verifier
Grader: link geometry aligns with the scaled aerial (centerline within 1 m), conflict areas set to circulating priority, inputs/routes exact; rerun queues within 10% of reference. Vissim COM is deregistered in the agent VM; grader opens the saved .inpx with its own COM session and reruns with fixed seeds.
Proposed partial credit
30% geometry; 20% conflict areas; 25% inputs/routes; 25% queue results.
R07.13.01.01-T07 Run a MATSim scenario with modified replanning and report modal split
Original instruction
Run MATSim 15 on fixtures/equil_mode/config.xml with 50 iterations, replanning strategies ChangeExpBeta 0.7, ReRoute 0.1, SubtourModeChoice 0.2 (car, pt, bike) and write output/summary.json with final modal split shares and average executed score.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- MATSim
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Grader checks the config diff (iterations and strategy weights exact) and reruns with the fixed seed: modal shares within 1 pt and score within 0.5%.
Proposed partial credit
30% config correct; 40% modal split; 30% score.
R07.13.01.01-T08 Convert an OSM extract to a SUMO network with specified options
Original instruction
Convert input/oldtown.osm.pbf-derived input/oldtown.osm into output/oldtown.net.xml with netconvert: remove pedestrian-only ways, guess traffic lights and ramps, join junctions within 15 m, and keep only passenger-accessible edges. Report edge and junction counts.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP04· 3D/spatial model- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
Grader runs netconvert with the reference options and compares: edge/junction counts within 1%, TLS count exact; reported counts match the produced net.
Proposed partial credit
60% network matches reference options; 40% counts reported.
R07.13.01.01-T09 Simulate GTFS bus service in SUMO and compute on-time performance
Original instruction
Import input/synth_city_gtfs.zip into fixtures/city.net.xml with gtfs2pt for the Tuesday 07:00-10:00 schedule, simulate with background traffic fixtures/bg.rou.xml, and compute on-time performance per route (-1/+5 min window at timepoints) in output/otp.csv.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO · Python
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Grader reruns: all 6 routes mapped, stop sequence matches GTFS; OTP per route within 2 pts of reference.
Proposed partial credit
30% GTFS mapping; 20% simulation runs; 50% OTP values.
R07.13.01.01-T10 Calibrate MATSim mode constants to a target modal split
Original instruction
In fixtures/smallcity_matsim/, adjust the scoring ASCs for pt, bike and walk (car fixed at 0) so that after 100 iterations the modal split matches input/target_split.csv within 2 pts per mode, without changing other scoring parameters. Save output/config_calibrated.xml.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- MATSim
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
long· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
Grader reruns with 2 seeds: each mode within 2 pts in both; config diff limited to the three ASCs.
Proposed partial credit
15% only ASCs changed; 85% split error (pro-rata to 5 pts).
R07.13.01.01-T11 Program a ring-barrier signal controller in Vissim from a timing sheet
Original instruction
In fixtures/corridor_3sig.inpx (PTV Vissim), configure the ring-barrier controller at intersection 2 from input/timing_sheet.pdf (8 phases, min greens, max greens, yellow/all-red, ped phases, detectors on left turns) and run 1 hour. Export signal times table for intersection 2.
- Platforms
win· Windows- Applications
- PTV Vissim
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP08· App/system state- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Grader reads the RBC settings from the saved file: phase parameters and detector assignments exact; signal times table from rerun matches reference (green start/end within 1 s). Vissim COM is deregistered in the agent VM; grader opens the saved .inpx with its own COM session and reruns with fixed seeds.
Proposed partial credit
50% controller parameters; 20% detector assignment; 30% signal times.
R07.13.01.01-T12 Compare corridor emissions with 0% and 30% EV share
Original instruction
Using fixtures/ringroad/ (SUMO), run the scenario with the HBEFA4 passenger class mix in input/fleet_2026.csv and again with 30% of passenger cars replaced by zero-emission vehicles. Write output/emissions.csv with total CO2 (t) and NOx (kg) for both runs.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Grader reruns both with seed 7: vType distributions match the spec; totals within 1%.
Proposed partial credit
40% vehicle mixes correct; 60% emission totals.
R07.13.01.01-T13 Place loop detectors in netedit and drive actuated control via TraCI
Original instruction
In fixtures/actuated/ place E1 detectors 30 m upstream of each stop line with netedit, then write output/actuated.py using TraCI to extend greens in 3 s steps up to max green per input/actuation.json. Show average delay at least 10% lower than the fixed-time baseline.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO · netedit · TraCI
- Interfaces
desk· Desktop GUIterm· Terminal- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clickdrag· Drag & dropcode· Code & commands- Work product
WP06· Code/engineering- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Grader: detectors at 30+/-2 m on all approach lanes; controller respects min/max greens (TLS log); 3-seed mean delay reduction >=10%.
Proposed partial credit
25% detectors placed; 35% control logic respects constraints; 40% delay reduction.
R07.13.01.01-T14 Estimate evacuation clearance time with a contraflow lane
Original instruction
In fixtures/coastal_evac/ (SUMO), load the evacuation demand input/evac_demand.rou.xml and compute zone clearance time (last vehicle exiting the zone) for the base network and with the inbound lane of the highway reversed (contraflow). Write both times to output/clearance.json.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Grader checks the contraflow network edit (lane direction and connections) and reruns both with 3 seeds: clearance times within 2%.
Proposed partial credit
30% contraflow edit valid; 35% base clearance; 35% contraflow clearance.
R07.13.01.01-T15 Calibrate a Vissim freeway corridor to travel times and queues
Original instruction
In fixtures/i5_corridor.inpx (PTV Vissim), calibrate Wiedemann 99 parameters (CC0-CC2 per link behavior) and desired speed distributions within input/bounds.xlsx so simulated travel times (3 segments) are within 10% and max queue at the on-ramp within 15% of input/field_data.xlsx averaged over 5 seeds.
- Platforms
win· Windows- Applications
- PTV Vissim · Excel
- Interfaces
desk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & formsvis· Visual judgment- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro applong· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
Grader reruns 5 seeds: segment travel-time errors and queue error vs field data; parameters within bounds; network geometry unchanged. Vissim COM is deregistered in the agent VM; grader opens the saved .inpx with its own COM session and reruns with fixed seeds.
Proposed partial credit
15% bounds respected; 50% travel-time targets; 35% queue target.
R07.13.01.01-T16 Measure pedestrian crossing delay in SUMO
Original instruction
In fixtures/crosswalk/ (SUMO, striping model), add the pedestrian flows from input/peds.csv on both crosswalks, run 1 hour and write mean and max pedestrian waiting time at the crossing to output/peds.json.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
Grader reruns with seed 1: flows match the CSV; mean wait within 2% and max within 5%.
Proposed partial credit
40% flows defined; 60% waiting statistics.
R07.13.01.01-T17 Run dynamic user equilibrium to a relative-gap target
Original instruction
Run duaIterate.py on fixtures/region/ with the OD matrix input/od_am.xml until the relative gap is below 0.5% (max 50 iterations) and write output/due.json with iterations used, final relative gap and total vehicle-hours.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO · Python
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Grader re-evaluates the final route set: relative gap <0.5% computed from the final iteration's travel times; vehicle-hours within 1% of rerun.
Proposed partial credit
40% convergence reached; 30% gap computed correctly; 30% vehicle-hours.
R07.13.01.01-T18 Find the gridlock junction by watching sumo-gui
Original instruction
Open fixtures/gridlock/scenario.sumocfg in sumo-gui and run it. A gridlock forms during the simulation. Identify the junction where the blocking queue spillback first closes a loop and the simulation time (s) it starts, and write both to output/answer.json.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
nav· Pan/zoom/scrollvis· Visual judgment- Work product
WP12· Answer/decision- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Grader reference from offline analysis of the deterministic run: junction id exact, time within 30 s. The agent has no terminal (sumo CLI and TraCI unavailable) and extra outputs are disabled in the provided config (hash checked), so the junction must be found by watching sumo-gui.
Proposed partial credit
60% junction; 40% time.
R07.13.01.01-T19 Evaluate a bus-only lane on an arterial
Original instruction
In fixtures/arterial_bus/ (SUMO), convert the curb lane of edges listed in input/buslane_edges.txt to bus-only (allow bus), rerun, and write output/buslane.json with average bus travel time end-to-end and average car delay for base and scenario.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Grader: lane permissions changed only on listed edges; reruns 3 seeds: values within 2%.
Proposed partial credit
30% correct lane edit; 35% bus times; 35% car delays.
R07.13.01.01-T20 Quantify freeway capacity gain from CACC platooning
Original instruction
In fixtures/freeway_cacc/ (SUMO), run the merge-bottleneck scenario with 0%, 25% and 50% of cars using the CACC car-following model (parameters in input/cacc.json) and measure bottleneck discharge flow (veh/h) from the downstream detector. Write output/capacity.csv.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- SUMO
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP07· Simulation result- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Grader reruns 3 seeds per penetration: discharge flows within 3%; vType settings exact.
Proposed partial credit
30% vType definitions; 70% discharge flows (three cases).
Mapped research references
Existing benchmark-to-class links from the research catalog. They are not record-level annotations, executions or product endorsements.