We are testing live tracking, aiming to figure out how to use it as a training aid. All test rides for the camp are grouped. It was a last minute invitation to test at the camp due to positive feedback at two earlier presentations. The late decision precluded preparation.
Friday was a pursuit with a morning session swapping handsets between riders. Live data was provided from one handset. Two riders were tracked on the way home. The afternoon was a repeat of the morning with live, corresponding uploaded data and earlier uploaded data from one handset and live, corresponding uploaded data and earlier uploaded data from a second handset. We missed some change overs because recovering a handset and passing it to the next rider breaks the flow. It also requires records of who is riding when. Handsets aren't expensive so a handset per rider is the way forward. We are purchasing ten more for future testing. You don't want to interfere with the flow of the session therefore starting the handset early in the day and stopping at the end is better. Remote control of the app through a web interface by a third party would be better again. This would enable battery management and sessions to be broken down into separate activities live without bothering the athlete.
Dave Martin mentioned it would be useful to use the tracking data for recording times, particularly where there are a lot of riders. We have a three second resolution in the live data and one second in the on device My Tracks record so timing will not be high resolution but interpolation may get timing a bit better than one second. Using the one second data, pursuit times for a few riders were estimated as:-
|Lap||Rider 2||Rider 4||Rider 5||Rider 7|
|Measured Distance (km)||3.36||3.30||3.34||3.28|
Rider 3 recorded power data on the ride back from Queanbeyan and another cyclist tracked position only.
Saturday Hill Climb
On the morning travel to Black mountain Rider 17 and Rider 14 carried a handset and Rider 14 provided power data from a PCS 7. Two riders in the morning group carried a handset up Black Mountain and two in the afternoon. Another rider didn't want to carry the extra 136 grams, prompting the question of whether that was due to the idea of extra weight or whether 136 grams really is too much? The hill climb needs a lot of staff at the top to get individual arrival times in case of bunching so GPS timing here might be useful. In this case it was a rolling start so knowledge of the timing start point is required for a timing estimate. The morning riders were Rider 17 finishing eigth and Rider 13 finishing sixth. Neither had a PCS 7 but another rider did so I tried pairing both units to Rider14's PCS 7 and you will see the same power data on both graphs. When discussing live tracking some people express the view that live data shouldn't be shared as it gives your opponent an advantage, however if PCS 7 telemetry is turned on anyone who wants can pick it up if they are close enough. The field strung out and once the distance was too great the radio signal and therefore power data was lost. On this occasion there was a single PCS 7 so it was easy to connect to the right one but the current My Tracks pairing algorithm of connect to the first unit it can find will be unsuitable in an environment of multiple PCS 7s. Also there is also only 1000 possible ANT + addresses for a PCS 7 so in a big event its likely multiple units will use the same address and interfere with each other.
One volunteer and Rider 3 carried the handset in the afternoon with Rider 3 recording power data. The afternoon riders came in close together with Rider 3 fifth. On the descent the riders had to get to the bottom as quickly as possible but they faced culling if they crossed the white line or crashed.
On Sunday there was a three hour race to Yass and another back in the rain with some dirt sections. Riders were too spread out for a peloton. These are expensive bikes which I thought were fragile and while they didn't look good on ruts or in the mud with their thin racing tyres, nothing broke. Going downhill fast in the mud looked scary and I saw one rider drop back from her more courageous colleagues in fear. No one complained though, and the winner from Tasmania said. "It suits me. Its like home".
We expected loss of mobile reception along this route so Umran spent Saturday improving the logic for dealing with the loss of connectivity. There wasn't time to install this on Rider 3's handset on Sunday morning. Rider 3's handset got a server error shortly after setting off so the app stopped sending. This meant loss of live data but the data is still captured on the handset by My Tracks. Rider 3 forgot her PCS 7, so I picked it up and delivered it to her half way to Yass. This allowed us to get live power data for most of the way back, losing the live feed when the mapmytracks server shut down for maintenance. Again the data was still available on the handset. Rider 14 carried the Xperia ARC handset with the software upgrade and it provided live on the way out until we lost the GPS signal and all the way back. The power data is good but the GPS data was patchy so we checked the live data against the handset data file on the way out and back giving the same result and showing the problem is this handset's poor quality GPS. It provides an example of the necessity to test handsets. This is a much newer and more expensive handset than the X10 mini pro, but for this application, inferior. For anyone who's reading this in Australia and wants an X10 mini pro they are being advertised at Target for $99 sim locked to Telstra between 24 Nov and 30 Nov 2011. From an observers point of view the loss of mobile reception is disconcerting because they see the rider stationary for a long period followed by very fast travel when the mobile reception returns. The new software dealt better with loss of connectivity so I'll upgrade the market app in the next few days.
There was a 15 minute break at Yass which allowed for a battery change. This takes about 3 minutes on the X10 mini pro and probably double that on the Xperia ARC as it has a slow boot. Battery changes are quite practical in a break but it ought to be a support person that does it as the athletes are too tired to fiddle. This three hour event each way provided a good opportunity to test battery life. Both handsets lasted the distance both ways. The Xperia X10 mini pro had at most another half hour of battery life remaining and the Xperia ARC was a bit below half in each direction.
Live tracking would be more useful with more riders so Dave Martin suggested at the Friday night debrief installing the Iphone app to the 13 Iphone owners and our app to the Android owner. Nobody did, perhaps because everyone is already tired and its too much bother, providing further evidence that tracking has to be low effort to get buy in and people need to be familiar with the technology prior to an event. We've seen teams tracked at the past two Tour de France races, but not the field. We need to see the full field on the one map to understand an event so it would be great to do an event with universal tracking. This is a weakness in the SRM model where only riders with SRM telemetry can connect to SRM Live. Its more valuable for everyone to see the data from other riders, even if its only position for those without power meters, so live tracking systems ought to work with a wide range of equipment. The majority of the riders on the camp already had a suitable smart phone, so we could have done it. Also, even at this level there were only 3 out of the 21 riders with PCS 7s. Not everyone with a power meter had a PCS 7 so widespread use also needs My Tracks to connect to more sensors including directly to power meters which Google is working on. GPS accuracy is currently critical. A handset with poor GPS like, the Xperia ARC or an environment unsuitable for GPS will give poor speed and distance accuracy. One solution would be to use odometry for the speed and distance measurements and the GPS, for location only. The mapmytracks server goes down in the middle of every night in Britain, where the developers are based, for an hour or so of maintenance. This probably doesn't bother people there but it's a significant inconvenience in Australia, where it is mid afternoon, even though we managed to record across the shut down with Saturday'software upgrade.
- We need to track everyone in an event on the same map to understand what is going on.
- Handsets need to be tested prior to use as newer and more expensive is not always better.
- Handsets should be associated with riders not swapped between riders or associated with bikes.
- Battery life is not an issue up to 3 hours with the X10 mini pro but needs thought for longer events.
- Riders need to be familiar with using tracking prior to an event.
- Effort to set up and go needs to be minimal, ideally nil. Its not practical for a rider to manage the technology during an event.
- Remote control of tracking would be useful for separating activities and managing battery life.Tracking systems need to connect to a variety of devices including directly to power meters and heart rate monitors.
- The daily mapmytracks server shut down is a significant inconvenience.