# Generated Data

The KITTI data does not provide all the data types we would like to demonstrate. Specificially data
related how it makes predictions about objects and makes plans based on those predictions.

For the sake of demonstration purposes we have generated additional from the base KITTI data set to
showcase a features to show how XVIZ can be used in a more complete data environment.

The KITTI converter has an option `--fake-streams` that can be used to output additional XVIZ stream
described below.

## Vehicle trajectory path

The KITTI data does not provide any planning information which is common in systems. However we can
use the stored location information to generate a trajectory to preview as though it was the planned
trajectory.

The `/vehicle/trajector` is generated by looking into the future known position of he vehicle and
generating a polyline to show the path that would be taken.

## Variables

The planning process of a system deals with the future timeline in order to make decisions now.
Unlike **time_series** data in XVIZ, **variable** provide the mechanism for capturing this future
timeline metric information for a given timestamp.

Multiple streams with the `/motion_planning` prefix will be generated to show additional plot
capabilties.

## Objects future_instances

Similar to how **variables** provide you the ability to capture metrics over future timeline,
**future_instances** provide the same ability for **primitive** objects.

This data always generated in the KITTI stream `/tracklets/objects/futures'. This can be seen in the
**streetscape.gl**
[sample application](https://github.com/uber/streetscape.gl/tree/master/docs/get-started/starter-kit.md)
using the slider under the playback control in the lower right.