Fig 1. A half-height cabinet was required to fit the hardware.

 

The Speedway Tracking System tracked 4 bikes around the oval dirt course and produced real time 3D coordinates for each rider. Riders were identified based on the colour of their helmet hood.

Following a successful indoor prototype, I created an 9 camera Speedway race tracking system in the spring and summer of 2003. At the time the hardware was quite clunky – Pentium 4s were cutting edge and due to the extreme conditions at the track (i.e. dirt, dust and rocks flying everywhere!) CCTV enclosure-protected PAL cameras was about as high end as we could go on the video ingest side.

Some of the key lessons of this prototype were:

  • Outdoor environments are nothing like indoor environments in terms of difficulty!
  • The more challenging the conditions, the more your camera placement matters
  • Tracking systems must be designed with graceful degradation in mind, as in the real world data degradation is inevitable.
Fig 2. An example race tracked by the system – in this undistorted system display the riders are identified by coloured blobs, and some of the dirt and dust is filtered out of the image

These days one could recreate a system like this in the space of a shoe box for a fraction of the cost, but back in 2003 the prototype required a half-height cabinet due to the form factor of the video ingest hardware and the computers.

Specification:

  • Three Pentium 4 class computers
  • 9 QPAL camera feeds in CCTV enclosures
  • Automatic tracking of Speedway races
  • Automatic identification of riders

Speedway Tracking System in Pictures – Clickable Gallery

Indoor prototype The system was initially developed and trialed indoors with 3 video cameras and an remote controlled miniature vehicles
Indoor prototype Besides just proving out the general concept, the indoor prototype also allowed for end-to-end testing of the downstream architecture.
Migrating system to the real world Once the prototype had been successfully proven out, the work to migrate the technology to the real world began. Here coax cable spools are being assembled for delivery to the track.
Migrating system to the real world In migrating the system to the real world,the system also required more robust housing for the electronics and computers.
Setting up at the stadium Mounting, testing and calibrating cameras at the track required the use of a cherry picker.
Setting up at the stadium The coaxial cables were also vulnerable to physical damage, and cable routing required some forethought.
Setting up at the stadium Mounting sensors often required working around the speedway track's own infrastructure. Mounting a camera close to a bright spotlight is not generally recommended!
Weatherproof CCTV enclosure Weatherproofing the tracking system PAL cameras was straightforward thanks to off-the-shelf CCTV enclosures
Low light effects For evening events the system had to cope with longer exposure times, resulting in motion blur.
Starting line A dedicated camera was mounted for automatically identifying riders based on their helmet hood colours at the start line.
Speedway bikes and dust Another issue the system had to cope with was the (sometimes very significant) amount of dust that the bikes generated in the bends. Note the tracking camera on the lamp post.
Tracking system For a sense of scale, I am standing next to the physical tracking system inside the control tower (pretty slim system for 2003!)