How does the maximum tag speed of the DWM3000 differ from that of the DWM1000? The maximum tracking speed of a DWM1000 tag is 5 meters per second. I think I read somewhere the DWM3000 was about 7 meters per second, but I cannot remember where I saw that. Thank you!
I’m not sure where you’re getting those numbers from. The maximum speed will depend on the radio settings used but 5 m/s is certainly not the upper limit.
We have sucessfully tracked a car at 160 km/h using a DW1000 based system. Since the car wasn’t heading directly at the anchor the relative speeds would be slightly lower but still in the region of ~40 m/s. In that situation the limit wasn’t the tracking system it was space and safety, going that sort of speed indoors you tend to run out of building very quickly.
1 Like
At those higher speeds, does position accuracy suffer at all?
That is a very tricky question to answer.
It doesn’t look like it does.
A plot of position over time doesn’t look any more noisy when going faster.
If we overlay position from UWB and from some other source (e.g. RTK GPS if outside) then the lines look to be just as close together at high speed as at low speed with the largest difference being due to the different antenna locations.
But at high speed any wobbles will be more streched out and so to the human eye it’ll look much the same even if it is actaully worse.
When we looked at the difference in position between GPS and UWB for individual times that did increase with speed. But the error was always in the direction to travel which indicates that most of what we were seeing was a slight difference in when the position was being measured by the two systems.
The accuracy we get is still in the order of a few cm (we typically get 2-3 cm static accuracy) but trying to measure accuracy at high speeds to within that sort of tollerance gets tricky. I don’t think we get any meaningful degradation but if it was dropping from 3 cm to 5 cm it would be difficult to detect.
It should be noted that I’m not using any of the qurvo supplied ranging and positioning systems, our system was designed from the ground up to give accurate positions and high update rates for things moving quickly. It does this by trading off a lot of the flexibility found in other systems. So this is very much what is possible with the devices if you try rather than what the standard supplied code can do.
One thing to keep in mind at high speeds is that raw measurment rate matters a lot. Any TWR system will be measuring one range at a time. It will then calculate position using those ranges, normally with the fundamental assumption that the tag hasn’t moved over that time period. As a rough estimate without accounting for this you are going to get an additional position error of the same magnitude as the distance moved while those ranges were being measured.
So if you move at 10 m/s and take 10ms to make all your range measurments then that’s going to give you an additional position error of 5-10 cm.
I get around this two different ways. Firstly by measuring the ranges very quickly so you don’t have much time to move (2400 DS-TWR results per second). Secondly by tracking when each range was measured and applying a filtering and smoothing system that allows me to adjust the ranges to correct for the time differences. This way the position calculation is fed what the ranges would have been if they were measured at the same time assuming a constant velocity over the measurment period. If the measurment period is short you have to be pulling several g for the change in velocity over that period to be significant (10ms period and 1g = 0.1m/s change).
In theory you could account for the movement when calculating the position. I did try a position calculation system where the current speed and heading were also calculated and the ranges were automatically adjusted to allow for this based on the relative positions and measurment times. I couldn’t get it to work well, it was a lot slower with no improvment in accuracy. But in theory it should work.
Thank you for the thorough answer, Andy! Currently, we are looking walking speeds or maybe a bit faster. I believe the maximum we may look at in the future is 30mph.
Andy, I have a question on the speed and max range. We have developed a prototype with DW3210 for a relatively high speed application (120Km/h max) while we want to keep the distance measurement range larger than 200 meters (with PA and LNA). But we found when speed is increasing, the range will be decreased (less success distance measurements), even in a speed of 15Km/h, the distance decrease a lot (75 meters). Is it normal? Or the protocol design matters? We are optimizing the RF circuit design. But we are not confident, the improvement would be as expected.
Our market is automotive so even the most basic test you’re looking at > 30 km/h, if range was halving at 15 km/h I’d expect we would have noticed.
We’ve tested at ~160 km/h and not seen a significant drop in range. The system performance in terms of anchors used remained the same at those speeds as it did at lower speeds in the same area.
We do use the shortest length packets we can, that was done for update rate reasons but may have an impact. High update rates also help for high speeds, most position calculation systems assume all ranges were measured at the same time. If you are using TWR then that’s not the case. You can compensate for this but getting them as close together means less to compensate for.
The other thing to check is that it really is the speed causing the issue. Excessive vibration could mess with the clock phase noise which would significantly impact the performance.
Andy, thanks for the reply. The project is still in its early stage. We are doing more testing on the high-speed performance. Another question is the impact of rain/snow, have you performed any tests in bad weather conditions?
The only weather related issue we’ve hit was when we had an assebly processes issue that compromised the waterproofing on a batch.
Most units are indoors (if you are outside GPS is a far simpler option) but we have had a few outdoor sites and some of the indoor sites include fake rain. When assembled correctly to be waterproof we’ve not hit any weather related issues.
That said I’d expect performace to suffer greatly if covered with a layer of snow.