I would like to discuss an application idea with developers here who are keen to develop the world's first motorsport performance analysis tool for the Nokia Internet Tablet.

Seeing that the Internet Tablet can be paired with a GPS device (or use the built-in one on the N800), here's what I have in mind

Seems to me that it would be possible to get the N800 running as a logging tool for trackdays when paired with a GPS software, if someone would be able to cobble together some code to log and analyze the log data.

The way I see it, using the logs and the correct formulas, it would be possible to have the N800 compute the rate of acceleration/deceleration/speed at any given point, average speed between two or more points that can be user-defined, and time taken between two or more user-defined points (for lap timing and sector timing). User should also be able to define start/stop lines (which may or may not be the same coordinates), pre- or post-event, and every pass across that virtual line counts as one lap.

Even with most commonly available GPS receivers, this should be good for up to +/- 0.3s of accuracy, which isn't too bad if the software isn't too costly to begin with.

It would also be useful if we could use the USB Host mode to log specific data from the car (oil temp, water temp, MAP, etc) alongside the GPS logs.

I believe the processor on the N800/N810 should also be able to display this information in real-time, i.e. best lap time in current session, average lap time, Vmax, warning indicators (if tapping to the ECU was possible, we could see oil temp, water temp, rpm, etc)

Versions:

The first version could be done to only log GPS data and then analyze the data post-logging. After the logs have been saved, the app can display the various paths (lines) taken by the car, and the user designate the start/stop line. Version 1 would only allow 1 start/stop line to be defined. Based on the defined start/stop range coordinates (app defines it as a line 25m perpendicular to the direction of travel in the first lap

Post-event analysis should show:

Lap Lap Time
1 1m 20.4s
2 1m 18.7s
3 1m 19.3s
4 1m 22.6s
5 1m 15.2s Best Lap
6 1m 16.5s

Up to 30 laps can be recorded in a single file.

User can then save the log file for later analysis

Later versions could offer

real-time display
- showing current lap time (left pane, large numbers, minutes on top, seconds below)
- best lap time (right pane, top 1/3 of screen, medium-sized numbers)
- last 5 lap times (right pane, bottom 2/3 of screen, in small font

offline analysis of sector time
- time taken to travel between two user-defined points on the track
- user can define up to 99 points
- compare time taken between point 1 and point 2, point 2 and point 3, point 5 and point 8, etc -- user selects the points to reference, and system computes based on GPS coordinates

lap compare
- user selects the laps to select, up to 5 laps.
- On selecting the laps to be compared, user clicks 'Compare' and it will show overlays of one lap over another, in different colours, 2 pixels wide)
- When user clicks on 'Play', a large dot will appear on the paths of each lap, indicating position, speed (displayed on a pane on the right), and acceleration (+ = acceleration, - = deceleration), so that the user can see the cause of the time differences between laps.
- User should be able to control playback speed, from 100% (real-time) to 25%, in 10% increments

Connectivity to ODB-II
- generic drivers similar to Carman can be used to gather data from the ECU on engine speed (rpm), coolant temp, ignition timing, etc and logged against the GPS logs, so that users can study behaviour of the car at any given position of a lap
- screen can be made to blink different colours depending on the warning type, e.g. if oil temp > 140 deg C, flash screen green; if rpm > 7000 rpm, flash screen red -- or use a pane on the right to flash.

If there are any developers who are keen to work on this with me, I would be happy to do so, and we can either release it as freeware or commercial.

I know a number of people who would be keen to use such a software, and I'm sure some of us here are petrolheads in addition to NiT zealots

Cheers

Kenneth

People I think would be able to pull this off:

The guy(s) who did Carman, and Maemo Mapper

handful?

hi guys, ghoomk...

Actualy is guys =) and we didn't do maemo mapper .. but we did this :



Just like maemo mapper, with navigation and more.. integrated in canola (what we call Canola Car Edition...


But just my 2 cents: I think the idea is great. As concept / interaction designer for the projects (carman / canola ) we wanted to provide this, I don't know if the version is going public soon, but more or less what you say we released at LGPL thus you could use it to build your software =) that's the beauty of open source.

The plans for carman was exacly like that.. you have a "rich" ui that is used just for in car , tuning lovers satisfaction, and the GTK UI where you can create all kinds of reports, logging with or without GPS to generate several kinds od data.

One of the concepts we implemented was to be able not only to log everything from you but also from your friend, with Carman's communicating via internet .. thus you can see your friends info on the screen..and log them also to compare with yours.

I think the combination of OBD2 with GPS very powerful...So if you need help (actually we don't have time.. but we can always try to help) contact us in the channels in freenode.. or via email!

BR

I've thought about this too. I've got my N800 mounted to the handlebars of my motorcycle with Maemo mapper running too.

I actually know the guys who provide the GPS systems they use in NASCAR over here in the USA. To get the accuracy they wanted, they combined a GPS system with an inertial measurement system. Seems that GPS suffered from occasional reflection off of other cars.

Is there simple snippets of code somewhere to show GPS logging? I'd like to play with it. I have a track day scheduled in about a month down here in Houston.

David

The problem for this is GPS accuracy. It just isn't good enough, and won't be soon. You'll certainly need 5Hz sampling, but the position errors will be a problem. GPS is only good to about 30 feet accuracy, and sometimes less. Reflected signals can cause much larger errors. There is survey-grade differential GPS available, but it's expensive and not very portable.

@sgosnell : I'm using a 5Mhz receiver, not the one in the N810 as I understand that one is 1 Mhz

@ handful/marcelo : I would be happy to take this offline with you, just send me your email address and if you are keen to take up this project, we could take this forward. I know you guys have done stuff like this, and if we take this forward, we can develop the logging component and basic lap timing as phase 1, then introducing the rest of the features as we go along. I will leave it to your discretion on whether you want to offer this as freeware, commercial, shareware or donationware. There is a fairly large market of amateur motorsport enthusiasts who would balk at spending over USD800 on a unit such as the RaceLogic Performance Box (and who don't need the accelerometers), and the Nokia Internet Tablet would make an excellent alternative offering lap timing, analysis (which you could offer as a Pro version), GPS navigation (good to be able to get home after track day), music and video jukebox and mobile internet device.

Integration to Carman would allow users to diagnose and reset DTCs without needing to lug around a separate device. This integration could be used to display real-time graphs and warning indicators.

I have a Tactrix cable (http://www.tactrix.com) that I use for datalogging ECU data from my track Mitsubishi Lancer Evolution VIII MR and use Mitsulogger (http://www.aktivematrix.com)

When this project is complete, you guys can look into selling hardware, software and accessory bundles, and we can test compatibility across multiple vehicle makes and I can assist in the promotion of this setup.

Let me know -- As a product and launch manager, I would like to offer you guys my years of experience in product development (from a requirements perspective) and marketing if it would help take this project forward.

I disagree with you on this point (I think). Although GPS's absolute accuracy is not going to be enough (certainly with a single GPS and no additional DGPS information), which is why surveyors must use something more high-tech, we're talking about relative changes here.

For this, my understanding is that GPS not too inaccurate. Certainly there will be the standard problems with poor signal, but the more immediate problem is that you can probably only get data out at 1Hz (i.e. one reading per second). This will work for gross measurements, but for acceleration calculation you'll need a higher sampling rate, which means a more expensive GPS system that is able to output at higher frequency.

Which goes back to my earlier point that a 5mhz receiver should be used for optimal results. These do not cost much these days and certainly cost less than a semi-professional datalogging tool.

I believe the internal GPS on the N810 will work, but is sub-optimal

Forget GPS and use WII remote. There is already many projects + maemo garage projects Py Axel Wii (http://pyaxelwii.garage.maemo.org/)

Well you could use gps for accuring and sync location coordinates when you are standing still (GPS accuracy as it best). I believe this could be great feature for maemo mapper as well. N800 users wouldn't even need gps because they could just point out start position at map layout. Thats assuming they know where they are.

Don't think it'd work that well. DJS spoke of the NASCAR system using GPS and accelerometers. The accelerometers get you short time-scale, and the GPS long time-scale. You can't use (consumer-grade, wiimote-ish) accelerometers alone for long time-scales because of the double integration. Basically, your velocity error increases with time, and then your position error increases as the square of time, so you need GPS to correct it.