In post-processing it is possible to run the inertial (and GPS) data backwards in time as well as forwards. When running forwards the end of a tunnel has the largest drift on it whereas when running backwards this tunnel has not yet been seen in the data so there is no drift. Combining the forwards and backwards data together has a clear benefit to drift.
There are other benefits too. Multipath from GPS is reduced. Velocity is improved since each time epoch is a combination of two (almost) independent measurements – the forwards and the backwards data. Heading, pitch and roll also benefit from these improvements, resulting in a 30% improvement.
There are some specific places where combined processing really makes a difference. Some examples are included here.
When processing forwards in time, the drift at the end of tunnels is at a maximum. The blue line in this example shows a 5m drift at the end of the tunnel. The red line combines the forwards and backwards processing. At the end of the tunnel the backwards processing, which has no drift, is heavily weighted and so the combined output is mostly from the backwards processing.
This data uses real-time OmniStar VBS corrections, which gives better than 1m positioning in open sky and is very easy to use and manage.
Combining forwards and backwards processing in urban conditions improves the position accuracy and reduces the jumps caused by multipath in the GPS data. It helps give more consistent results in difficult conditions.
In the urban conditions above the blue trace, processed in a forward direction only, is pulled down (south) by multipath from incorrect GPS. By combining forwards and reverse processing the RT is able to correct this multipath error and give results that clearly show driving on both sides of the road.
The Google images in this document have been created from a licensed version of Google Earth Pro. Copyright of the images remains with Google.