The data is organized as a series of timesteps. Each timestep starts with a row giving the local time in seconds followed by the number of cars that exist during the timestep.
[time (sec)] [n = number of cars in the timestep]
There follow n entries specifying the state of each car for the timestep. For each car, an entry consists of an id number for the car, its Cartesian x y z position, its direction as x and y components (nx and ny), its velocity in x and y directions, and its acceleration.
[ID] [x] [y] [z] [nx = cos(theta)] [ny = sin(theta)] [velocity * nx] [velocity * ny] [acceleration]
This data set consists of multiple lanes of northbound highway traffic. The data was created using a reconstruction process, and it exhibits jams, passing, and an HOV lane.Download (274 MB)