fundacio-i2cat / collisions_datasets

FEM-IOT-Col

Datasets of vehicle collisons in Barcelona. The datasets cointain information regarding the position of the vehicles in every step of the simulation, with a parameter determining if they are part of a collision or not.

There are two datasets: positions_v1 and positions_v2. The first one, emphasizes reckless drivers with a majority of back to front collisions. On the other hand, the second dataset is an upgrade with more diversity in the kinds of vehicles driving. Vehicles can also skip traffic lights, this way generating more lateral collisions in junctions.

Positions_v1

This dataset is generated with reckless vehicles. The majority of the collisions are front to back. The parameters that are in the dataset are the following ones:

time	Time instant of the simulation, in seconds
vehicle_id	Vehicle identifier. If a collision occurs is the one collisioning.
victim_id	Identifier of the vehicle victim if a collision has occured. If no collision has occurred de default value is -1.
latitude	Position of the vehicle in geographical coordinates: latitude.
longitude	Position of the vehicle in geographical coordinates: longitude
speed	Velocity of the vehicle in m/s
heading	Direction of the vehilce in respect to the north
collision	Boolean, it indicates if a collision has occurred or not: 1 collision ocurred 0 no collision ocurred

Positions_v2

This dataset has more diversity of vehicles than the previous one. Our objective was to increment the number of lateral collisions. To do it, the vehicles in the simulation can skip traffic lights. The diferent kinds of vehicles introduced in the simulation are the ones listed in the table below. Moreover, the simulation steps in this dataset are every 100ms in contrast to the 1s of the Positions_v1.

Vehicle Class	Length width height	mingap	acceleration	deceleration	emergency deceleration	Max speed	Speed Deviation
passeger	5 1.8 1.5	2.5	2.6	4.5	9	200	0.1
delivery	6.5 2.16 2.86	2.5	2.6	4.5	9	200	0.05
bus	12 2.5 3.4	2.5	1.2	4	7	85	0
motorcycle	2.2 0.9 1.5	2.5	6	10	10	200	0.1
moped	2.2 0.9 1.5	2.5	1.1	7	10	45	0.1

Moreover, two new parameters are beeing observed in the dataset.

1. Acceleration: it indicates the acceleration of the vehicle.
2. VehicleClass: it indicates the shape of the vehicle (passenger, bus, motorcycle, delivery or moped).

Positions_v3

This dataset is the same as the Postions_v2, however the simulation steps performed are every 1s as the Positions_v1 dataset.

Positions_v5

This dataset uses the same scenario as Positions_v2, but with larger routes for the vehilces and a longer simulation time.

Barcelona_v2

Dataset generated in a new and bigger scenario. It is located in Sant Gervasi, Barcelona. Dataset provided with a realistic approach, with the vehicles staying longer times in the simulation and circulating more in the principal highways.

Each simulation step of the simulation is every 1s.

The dataset is divided in two but they can be glued together again after downloading and decompressing.