AvocadoAC - Activity Classifier

Today I'm presenting work on the  AvocadoAC activity classifier application, for Android smart phones at the 5th Asia-Pacific Congress on Sports Technology .

Purpose of AvocadoAC

To continuously measure physical activity, classify activity by type and quantify activity intensity using the sensors of a smart phone.  Once known sports training activities are detected the specialist application MyTracks is invoked to record and analyse activities in more detail.

This information can be used to:

• Provide life-style information on total exercise, health and activity levels.
• Infer further lifestyle information by combining with other information sources, for example location data.
• Provide context to reduce data gathered and data management problems for sports tracking by only gathering high bandwidth data when it’s likely to be useful.

Background

The smart phone is an attractive platform for context-awareness because:

•  People carry mobile phones around in their day-to-day lives.
• Activity monitoring can run as a background application
• Smart phones are becoming ubiquitous.

Challenges

Battery life can not be significantly impacted.
Mobile phones are carried and used in a variety of ways.
Different model Android phones have different functionality. In particular some models disable the accelerometer when the screen is off.

Get the presentation . 
Try the AvocadoAC activity classifier.

Ken Taylor, U. A. Abdulla, R. J. N. Helmer, J. Lee, and I. Blanchonette. Activity classification with smart phones for sports activities. In APCST 2011: 5th Asia-Pacific Congress on Sports Technology, Procedia Engineering. RMIT University, Elsevier, August  2011. 162k View Download

Live Bicycle Tracking For The Masses

Tour De France Generates Interest

Live tracking was first available in the 2010 Tour De France and after I discovered it was done with a non public version of the open source Android application My Tracks and that the AIS cycling athletes  wanted live tracking I've looked at the problem more closely. In summary they used a HTC Android handset with the firmware specially modified to use the ANT+ functionality already available on the phone's WIFI chip.

For the 2011 tour live data for a few riders was provided by the SRM Live application. Telemetry is provided by an SRM module as shown in figure 1.

Figure 1 - The telemetry module is mounted behind the seat, has a GPS receiver and sends data via the mobile phone network

The SBS tour tracker was the obvious Tour De France audience tool for 2011 with versions available for PC, Android and iOS. The tour tracker showed live video, commentary and data, including data from SRM Live, but got mixed reviews with criticisms focusing on the implementation.  Another way to follow the action was to start with the twitter firehose using either #tdf or #tourdefrance then choose a few lower bandwidth tags, people or advertised sites to follow. My favourite site showed caricatures in profile with statistics and a commentary below. Mashups maximise the opportunity for creativity in presenting Tour De France data, providing information for expert commentary and, more broadly, presenting racing and training data in general. During this period of rapid innovation it is unlikely there will be one comprehensive solution.

Last Thursday, 4 August 2011, Dr David Martin presented an Australian Institute of Sport SMART TALK on "How the Institute helped former-AIS athlete Cadel Evans win the Tour de France". David mentioned he had the power data from the top three riders and that Cadel averaged 410 watts in the all important Stage 20 time trial. Live telemetry is experimental at the moment in big events and is available only for some riders but David expressed the view that, as for car racing where drivers are obliged to provide public telemetry, it is likely to be mandatory in future as live data can significantly improve the audience experience. David also mentioned that AIS athletes still don't have live tracking but they want it.

Live Cycle Tracking Applications

There are a wealth of sites tracking training data and some, like Endomondo and SportsTrackLive   that provide live tracking. They are evolving fast and are gradually incorporating more sensors and sophistication, for example the Zephyr heart rate monitor. People are mixing and matching data from different sites and sources, most often by importing and exporting data in one of the XML formats KML, GPX or TCX. Site and application specialisation is occurring, particularly with integration into social media applications but existing training sites are mostly pitched at providing an end to end solution. Likely niche specialisations are:-

• sensors  e.g. the Zephyr heart rate monitor or  SRM power meters
• data capture applications e.g.  My Tracks
• data repositories and data feeds e.g. Google fusion tables
• data presentation
• specialised analysis
• tailored advice from coaches and experts

Since the 2010 Tour, two handsets, have been released that support ANT + and My Tracks has added the functionality to communicate with the SRM PowerControl 7 . Though its still somewhat experimental, the components are there to make live bicycle power data widely available from consumer grade handsets as well as the  SRM Live application. A web service supplying live and archived data to upstream applications is an obvious service with presentation from multiple upstream applications like Endomondo and SportsTrackLive, but what follows? Is remote coaching practical?

Remote Coaching

I found the web coverage of the Tour De France greatly enhanced the viewing experience. There still seems a lot of opportunity for greater sophistication in data presentation but improvements are coming thick and fast.  Would adding a voice connection between the athlete and coach during training allow the coach to provide better advice remotely than than being there? Would it work for the athlete and would top coaches offer their services this way?

Update 9 Aug:
Thanks to David Martin for the advice that the 410 Watts mentioned for the stage 20 time trial is an estimate based on previously monitored fitness tests and time trials rather than data from the stage 20 time trial itself, which he hasn't reviewed.