Simulation data and code for validating two-dimensional time to collision for articulated vehicles

This dataset provides simulated data to validate two-dimensional Time-to-Collision measures proposed for articulated vehicles such as a tractor-semitrailer. These measures can detect both rear-end and sideswipe collisions, unlike the conventional TTC, which can only detect rear-end collisions. The files 'RearEnd.gif' and 'SideSwipe.gif' show visuals of a rear-end and sideswipe collisions, respectively.

The Simulation Data folder contains 30 CSV files, consisting of 15 rear-end and 15 sideswipe collisions generated randomly using a cut-in scenario in CARLA. Three different lengths of semitrailer are considered: 11m, 13m, and 15m. Out of 30 simulated scenarios, 10 scenarios are for each trailer length (5 each for sideswipe and rear-end collision). The attached file 'CarlaScenarioSimulation_Code.py' is used to create these collision scenarios.

Each CSV file follows a consistent naming structure. For example:
vehicle_data[Lateral/Rear]_XX_cY.csv
Lateral → sideswipe scenario
Rear → rear-end scenario
XX → semitrailer length in meters
cY → case index for that scenario

Each CSV file represents time-series simulation data. Each row corresponds to one simulation time step, and the columns provide vehicle states and kinematic information for the tractor-semitrailer and car.
x1, y1: Global coordinates of the tractor
x2, y2: Global coordinates of the semitrailer
x3, y3: Global coordinates of the car
vLx1, vLy1: Local-frame velocities of the tractor
vGx1, vGy1: Global-frame velocities of the tractor
vLx2, vLy2: Local-frame velocities of the semitrailer
vGx2, vGy2: Global-frame velocities of the semitrailer
vLx3, vLy3: Local-frame velocities of the car
vGx3, vGy3: Global-frame velocities of the car
yaw1, pitch1, roll1: Orientation angles of the tractor
yaw2, pitch2, roll2: Orientation angles of the semitrailer
yaw3, pitch3, roll3: Orientation angles of the car
delta: Steering input of the tractor–semitrailer system

The files '2D_TTC_predict_from_scenarios.m' and '2D_MTTC_predict_from_scenarios.m' read all scenarios and output 2D time-to-collision predictions at 1, 1.38, and 2 seconds for both versions for each of the 30 scenarios. The former uses the proposed formulation of 2D TTC that assumes constant speed and heading in the prediction horizon. The former uses the proposed formulation of 2D TTC that assumes constant acceleration and heading in the prediction horizon.

The cpp files cSpeed and cAcc are the C++ implementation of 2D_TTC and 2D_MTTC respectively for one of the scenarios to measure the computation time for one loop.