dbbarron

I used HLT for the first time at AX yesterday and had some great results.
First problem was video quality with iPhone 6+ - I think I have the shaky camera problem - genius bar appointment made. Likely Iphone issue, not mount or HLT.

On timing accuracy, from video analysis, the start and stop trigger points for HLT did wander a bit and caused some timing inaccuracies with HLT relative to official time.

The start point (average speed when crossing start trigger point was about 40mph) wandered by around +/- 0.25s around the actual start trigger.

The finish wandered around +/- 1second yielding an overall accuracy of HLTs time to be +/- 1.25s. This is way too inaccurate to fine tune or even compare AX runs which can be within tenths of each other.

I am using an XGPS160 which had a good signal and was providing fixes at 10Hz. Accuracy appeared high - 2.5m.

Is this a typical result?
Can anyone suggest how to improve it?
Would a VBOX solo with 20Hz update rate (but less accurate GPS) help?

Any thoughts appreciate.

Here's a video of a typical run from the day:

db

ProCoach

Why do you think the VBOX Sport is less accurate?

dbbarron

I believe XGPS160 is GLOSNASS and WAAS. Sees more birds!
XGPS160 rated 2.5m accuracy v. 5m for Vbox sport.

However, this is just a spec and real work results may differ.

Slakker

How are you determining your accuracy numbers? Have you exported to google earth to look at where the GPS point are? I would also look at your map to see your start/stop trigger points and ensure they have the correct heading.

Even with the 1hz internal iPhone GPS you should be getting much better accuracy. I assume you have looked at your lap list and verified it was using the Dual?

dbbarron

I have looked at both the accuracy graph and the GPS fixes in both HLT and in Google Earth.
There were a few run in the morning where HLT lost the DUAL connection and that is obvious from the data. I ignored these laps. For the most part, I show 10HZ GPS fixes with great accuracy. Each run appears to start and finish at the same point, but this would make sense. HLT thinks it is starting and ending the run at the same point. I also don't think we are seeing the actual GPS fixes, but HLTs interpolation of those fixes onto a route.

To determine my accuracy numbers I compared the first video frame from each of 5 runs within 1.5Hr in the afternoon. Since HLT thinks it is starting at the same point, I estimated the travel time between the various 'starting points' per the video frame. I then did the same for the last frame of the video which HLT believes in the finish.

Tomorrow I may create a start/finish trigger somewhere in my neighborhood and drive a few laps with the DUAL in different physical configurations and then same with the iPhone as GPS. Each should have the same first/last video frame as the other laps.

I've attached the Google Earth data for reference.

db

ProCoach

My experience does not measure yours. Period. HLT sucks.

Slakker

I'm still questioning whether there is actually an issue. The data points from the laps with the dual look pretty good. I'll look a little closer later as there The extrapolation comes from when it calculates you actually pass the finish line.

Was your club running a timer system? How did your the HLT lap times compare to it? With Dual, my times have been within hundredths of my transponder.

Also, did you check your heading on your triggers? Look under realtime and map then navigate to your AX track.

For something that sucks, it still has features that I wish Aim had. The ability to set your reference lap for your predictive +/- lap timing using any lap you've ever run on a track is pretty handy. That way you can always see how you are measuring up to your personal best.

It's also pretty nice that you can change your data, the track layout, sectors and your reference lap and overlay it on video after the fact. You can add your reference lap video to PIP and then remove it later if you choose.

I agree it doesn't have the reliability of the high dollar systems but for a "hobbyists app" it's pretty powerful.

Slakker

Actually, to check for accuracy, do a .csv export. The time stamp on the GPS readings will be WAY more accurate than the timing on the iPhone video.

ProCoach

This is true. I was only speaking of my personal experience at the track.

Slakker

It can definitely be frustrating. The fact that it is just an app and not an integrated system is its biggest drawback IMO. But Harold is a super nice guy who is passionate about his program and very easy to engage and work with. He'd make a good Rennlister.

dbbarron

I need to back up a bit. It appears that video start and end times do not directly correspond to the start/end trigger - the differential varies. Need to look closely at the video files outside of the HLT environment for more accuracy.

My club ran official time (lights) and other than one run there appeared to be a pretty solid relationship between HLT and the official lights, but there definitely was some variation in that offset of perhaps 0.5s - I don't have the official times yet. They should be posted soon and I should have written them down.

db

Slakker

HLT takes the GPS point before the trigger and the one after and then computes the distance of each from the trigger. It then uses that to extrapolate the exact time the trigger was crossed. The .csv file (you can open in excel) will give you the raw data used for that computation.

ir_fuel

An iPhone has a rolling shutter. You'll never have a decent image.

993Guy

Anything later than an iPhone 5 is pretty crappy for in car video. Something to do with image stabilization. There are threads over at the HLT forums on this. I used Harry's a while back, now running an AIM Solo DL and a GoPro. AIM is super easy, works every time, well worth the price.

dbbarron

I have a genius bar appointment this week. If I tap the side of the phone while recording, I get massive jumps in the image - Like the image stabilization's servos are not working properly (think the 6+ has this). If I do the same test on my daughters iPhone 6, no issue; my iPad Air, no issue, older iPhones, no issue. I'm going to try a few new devices at the apple store today.

Thus, I think it could be more than rolling shutter.

db