Simulation of 50 meter approach with environment

Simulation result

ATTENTION: This is not the data from the scanner of the UBER car. It is a simple approximation of a car approaching at 25 meters per second to a pedestrian with a bicycle.

Scanned with a virtual Velodyne HDL-64E. The output is the "raw" noisy scan without further calibration. Reflectivity has not been considered, but at a closer distances this should not affect the accuracy of this representation.

The scanner is mounted at a height of 2 meters above ground. This is a rough estimate of the mounting position where UBER has mounted the scanner.

Point cloud visualization

Below is a better view of the three point clouds in the video.

Distance to pedestrian: 40 meters

Distance to pedestrian: 25 meters

Distance to pedestrian: 12.5 meters


The environment is modeled after satellite images from Google Maps and views from Google Streetview.

Google Maps link to the location: 642 North Mill Avenue, Tempe, Arizona, USA

The models used are from

  • bicycle
Name: Mountain Bike [Unibike Mission 2010]
Author: hilux.
  • simple human figure
Name: Very Low Poly Human Basemesh
Author: tweediez.
Uni Salzburg