Some product ideas

  • Medicine reminder - lets you know when to take your medication (good for old folks)
  • Kid watcher - Keep track of your kids, alternatively let them know when to put on a sweater when the temperature drops (thermal sensors)
  • Thermometer sensor to constantly track kids with a fever. Could send notification to phone when a fever reaches threshold.
  • Sports tracker (fitness apps)
  • Swing analyser for golf swings etc. Also think football/rugby analysis with a dev board on/in a ball
  • Buzz notifications when long running tasks are complete (such as long compiles or db operations)
  • Pet wearables - put into a pet collar in a waterproof casing. I2C mic can tweet your dog barking etc


  • do you think the processor is fast enough for the golf swing analyzer?
  • @STIdeas You probably dont need a fast processor, just bomb the acellerometer data to a device via BLE in real time.
  • edited December 2014
    @STIdeas, we are able to do analysis for sports and other applications that are as "fast" as golf with no problems. 

    The accelerometer data can be streamed directly to your phone at 100 Hertz with high fidelity (up to 200 Hertz where you may start to lose packets because the BLE link saturates). If that's not fast enough, then you can simply log accelerometer data at up to 800 Hertz and download it for analysis on your phone later (for example, right after the golf swing).

    Usually we recommend that you gather as much data as possible and implement all your algorithms in software. 

  • brilliant.
  • As a general piece of advice, keep in mind some of the limitations of logging and streaming accelerometer data when designing your overall solution:

    - Streaming via BLE limited to <200hz from accelerometer due to bandwidth saturation
    - Logging limited to around 10-15 seconds of data at 800hz from the accelerometer

    We've run some fairly advanced algorithms on the MetaWear processor, but we almost always prototype in a high level programming language first on our desktops, then translate the code into firmware once a final implementation is decided on.

This discussion has been closed.