We have now completed first month of our CPS project. And already some interesting results and improvements.
1) Lookup algorithms.
Some more work has been made, but no new major findings. We have verified lookup on more sites. Error when knowing exact location of wifis is still in the 5m range depending on wifi density and no new algorithms sofar make a significant improvement to this. But adding a Kalman filter make the track look a lot better even if the mean and median error doesn’t improve. We will implement a Kalman filter in our Location-API app when we have time to make the tracks there nicer.
Our conclusion: Current CPS indoor lookup algorithms works well. We should add track lookup or history to improve further.Thus if we know the exact position of wifis, we should be able to get accurate indoor location.
2) Learning algorithms (main focus for our project).
To repeat, our goal is no manual learning and no calibration. All learning should be based on our crowd sourced data. We have now fixed so valid GPS positions lock down position for valid scans. The weight is depending on the GPS accuracy. This means good GPS positions with a good accuracy (below 5m) are heavily weighted and assumed correct. But weak GPS positions (~50m accuracy) that we may occasionally get indoor also has a weight that helps to make the track correct. This means our map is now fixed and we do not need to rotate or mirror any results as was required before. Improvement is shown in figure above. Blue dots are the track positions where we estimate the scan was made and red circles are showing where we think the wifis are. Figure left is our ground truth track. Center figure is current CPS API track. Right is our latest Indoor CPS track based on new algorithms. The improvement so far is significant. Errors are about half for this venue (15m median scan position error and 20m median wifi position error). We are also evaluating if and how much knowing a track (history) would help.
3) Our tool, the Location-API app, is updated and can now show track based on current CPS algorithms. Download and check yourselves. But to get really good indoor CPS with our app, we first need to recalculate the relevant area (tiles) in our database. Still more work required before it is time to update our database.
Now summer is here and we will continue our efforts. Stay tuned! Recent Posts
