Designed to integrate seamlessly with their VBOX product line of GPS/GNSS based data loggers the Racelogic VBOX Indoor Positioning System (yes VIPS, sorry, blame marketing for the name) allows the user to seamlessly transition from GNSS to UWB and back at speeds of over to 100 km/h while maintaining a constant 100 Hz position and velocity update rate.
Because all the calculations take place on the car on dedicated processors rather than taking place on a PC and then transferred back over a radio link the output has a low and constant output latency.
The system architecture allows for a large areas to be covered seamlessly. A 2-3 lane wide road could be covered for over 2 miles (in theory at least, if anyone has a 2 mile long tunnel we can use for testing then let me know.)
The system has been designed with ease of use a key consideration. Once installed in a building there is virtually no on car setup, just put the antenna on the roof and enter the separation between the GNSS and UWB antennas. The system will read everything else it needs to know from the building infrastructure.
The system also allows for temporary test setups, two people can get a basic setup covering 120 x 30 m with GPS alignment to allow indoor/outdoor transitions up and running in just over an hour. If you don’t need to go outside and are willing to sacrifice a few cm of accuracy then a quick setup can be done in around 20 minutes.
While the initial release is intended to be heavily targeted towards our existing automotive customer base using our VBOX product line the UWB system is fully independent and can output positions and velocities in latitude/longitude or meters at 100 Hz. The on car unit is a dome around 11 cm high and wide, is IP 67 rated and draws around 2 W of power.
For people who are interested in seeing it in action there will be a launch event in Bicester, Oxfordshire on the 23rd and 24th of this month. The system will also be on show at the ATE show in Novi, Michigan from the 22nd to the 24th this month.
OK that’s it, other than answering questions on this thread no more sales pitch.
In answer to some questions:
Does your system use TWR or TDOA?
TWR. We wanted very simple setup with no special cabling or placement restrictions on anchors being able to see a central point. That made TDoA, tricky to do well so we went with a TWR approach. Each anchors only requires power and doesn’t require line of sight to anything other than the car (when the car is using it).
Can it support multiple tags/cars?
This is the downside of the approach taken. For best performance we only support one car at a time. The initial market is brake / tyre testing where this isn’t a significant handicap. We do support 2 cars at a time, the performance drop is negligible in that configuration. I have tested 3 or 4 cars and the drop starts to becomes noticeable at that point and so that is not a supported configuration right now.
I have plans to fiddle some of the radio protocol we use and think I can get us up to 4-5 cars without any measurable drop in performance but that will be in a later release if it all works out.
Very few automotive tests involve more than 5 cars at a time so that is all we need. Obviously this does impact usability for other applications.
This does all depend on the dynamics of the cars, if we were to assume that things were moving slowly without large accelerations then we can cope with more cars and not notice any real difference. But the whole point of the system is that the car could at any time pull over 1 g in any direction, that requires a high raw data rate to capture events accurately.
What is the range between the Anchor and Tag you can achieve?
It depends a little on the details of the setup (heights, orientations etc…) but it’s around 60 m. We use up to 8 anchors at a time and typically set them up on a 25 to 30 m grid.
If you do go out of range of some anchors the system runs all the way down to 3 anchors (although speed and heading outputs stop if you go below 5) and when running with our VBOX we do a kalman filter to mix in inertial measurements which will fill in any short gaps in the data.
You have mentioned about “A 2-3 lane wide road could be covered for over 2 miles” - how many anchors do you need to cover a such tunnel?
2 miles = 3200 meters, at a 30 meter spacing is 107 anchors on each side so 214 total.
The system can cope with up to 249 anchors.
But I wouldn’t want to be the one getting the bill for a system that size
When stationary we will generally get a point cloud with a standard deviation in the 2-3 cm region.
The centre of this point cloud will normally be less than 5 cm from truth, often the error is around half of that but it does depend on geometry.
Repeatably is under 1 cm so in theory you could improve the absolute accuracy by mapping the errors and then applying a location dependent correction.
