I currently find myself developing my master´s thesis based on Decawave technology, providing 3D location for just 1 tag (I´d like to add some more in upcoming versions).
However, after having tested TREK1000 (comprises 3 EVK1000 anchors, and 1 more performing as a tag), it does not live up to my expectations in terms of precision. I have already achieved a minimum precision (standard deviation) of +/- 5 cm after filtering measurements 10 times and +/- 2.5 cm with a 40-time filter. I´d like to achive +/- 1cm but I can´t filter it anymore because of other requirements. Please, note that the offset error is not something I´m currently caring about.
Then, my question is, does DWM1000 improves EVK1000 results?. I have read that EVK1000 is already equipped with DWM1000 chip, however I don´t know if it is worthy to set DWM1000 up in this case.
Is there anyone who achieved a better precision? Does a 4th anchor provide more resolution? Apart from getting the right position from the 2 possible solutions obtained from trilateration equations.
Any answer is welcome
Thanks in advance.
Ignacio Valero
i.valero@alumnos.upm.es
Technical University of Madrid
The two both use the DW1000 IC for the UWB so the underlying radio is the same. The only differences I would expect to see would be due to antenna and environmental differences between the two modules. The antennas and presence of other objects close to the antennas can introduce some fairly significant orientation specific errors in the range measurements.
We have achieved standard deviation position errors in the order of 1-2 cm without any filtering / averaging, I know others have achieved better results than this. We did this using 8 anchors and a custom designed PCB/antenna using the DW1000 chip. Initially we used the DWM1000 but found the built in antenna introduced too much orientation dependency for our application.
Better anchor geometry, clearances around anchors and being careful with orientations is a quick and easy way to improve things.
Adding anchors will also improve things if you do it in the correct way. Yes with 4 anchors you can come up with multiple possible trilateration solutions (4 of them in theory). Or you can use all 4 at once using something like a least squares or kalman filtered system to find a best fit location for all of the measurements. Once you start doing that you can in theory add more anchors until you hit the required accuracy. In practice you do suffer from diminishing benefits for each extra one added and soon hit the point where the benefit isn’t normally worth the extra cost and processing required.