Saturday, 1 July 2017

LifeCloud lifelogging Sensing Camera 2017




Lyndsay Williams sensecam@gmail.com
Design of LifeCloud lifelogging Sensecam
Update: 25th November 2017


Bill Gates, Microsoft, explains what SenseCam is in Time Magazine.






Blackbird, Cambridge,  feeding on suet pellets captured  automatically by SenseCam


In March 2017 I was visited  by  Prof Cathal Gurrin, Dublin City University,  and colleagues from Flextronics, San Francisco  to design  a sensing camera system, for lifelogging applications for a startup company LifeCloud.  This is a small wearable camera that captures interesting events in the day,  original work here at Microsoft Research.   Software detects memorable moments, such as seeing friends, laughter or a change in view, and captures images in a few mS. The camera has a 170 degree wide angle lens, various environmental sensors, 14 hour battery life, blur reduction. Image search from the camera is provided by Google photos. Sensecam has proven to be useful re memory loss and dementia.
Here is a video from the BBC (4 mins) by James May ( BBC TopGear) visiting Alan Smeaton of Dublin University demonstrating the first Sensecam in 2008.

UPDATE July 2017. The LifeCloud company, of 8 people,  will be closing due to finance and management  issues.  Due to many requests for SenseCams from Universities and hospitals, Girton Labs and Lyndsay Williams are looking to progress the SenseCam under new management. 


We now wind the camera forward to 2017 and the latest Sensecam.

 I provided to LifeCloud a  sensing camera designs specification, working hardware and firmware.   One camera design was for preproduction trials  and one for production.   I decided to make  make the preproduction camera open source, hardware , firmware, case CAD design,  as per my blog post in 2009.



LW Sensecam 2017 Camera specification

  • 5M Pixel camera 
  • 170 degree wide angle lens f 2.8, fixed focus 4" to infinity
  • up to 1600 ISO light sensitivity
  • video,  image  and audio storage, jpeg, h264
  • smart sensing based on motion,  light change, audio,  thermopile, GPS  and other parameters
  • image stabilisation and blur reduction in firmware 
  • accelerometer, gyro, magnetic,  thermal and other analogue sensors
  • Bluetooth, wifi , HDMI, USB connectivity to phone and pc.
  • 12GB image storage, 500MB internal RAM , expandable up to 128GB
  • development firmware uses Linux and Python 
  • 3.7V Li Po rechargeable battery life 14 hours or 23 hours dependant on camera formfactor
  • weight of camera approx 90g
  • prototype case, black, white or translucent, size 74 x 50 x 22mm
  • neck lanyard and clip
  • $39 Bill of Materials , 100 off approx 
  • prototype based on Raspberry Pi Zero W 
5th July now with addition of working sensor module, comms not finished
  • power management of camera function 
  • thermopile distance temperature measurement  -40C to 125C 
  • 3 axis accelerometer 
  • 3 axis gyro
  • 3 axis magnetometer
  • humidity sensor
  • digital microphone
  • altimeter / pressure sensor
  • ambient light sensor 
  • loudspeaker
  • ​24 hours recording of above sensors in additional or instead of camera images.​
This all fits within the case as above.

All above is working, but not  yet the audio recording, external sensors,  (proven with 2004 SenseCam)  automatic wifi upload of images, this is done using the micro SD card.

Test video of sensecam 2017, 2nd April before anti blur software added. 



version  1 camera based on Raspberry Pi  April 2017





Next version, first test with new 3D Flashforge Finder printer





Edge detection in Raspberry Pi.




Button camera 

experimental circular Sensecam being printed by Flashforge 3D printer  recorded by another Sensecam, video here.




Early Lifecloud  design 


Above translucent camera with Raspberry Pi Zero W inside

current Linux GUI interface for camera 


Google image search results  for "spectacles Girton" in my Google Cloud.  Includes new Sensecam images.
Google found them on my kitchen table as below.





Original LifeCloud team, April 2017,  left to right, Martin Fridh ,[unknown],  Jim Myrick (white coat), Lyndsay Williams, Gordon Bell, Rami Albatal,  Cathal Gurrin (wearing SenseCam)


My  Cambridge engineering lab





Example image from my garden Sensecam


16th June,  open source design by  Dublin City University, by Cathal Gurrin,  the OpenInitiative  openi https://openi.co/

I plan to work with Cathal  Gurrin re this, contact me Lyndsay Williams, sensecam@gmail.com re design files for BOM, firmware and CAD case, or to arrange  a public share online.






Future production design

 For a built of 100 cameras, a Raspberry pi Zero W is economical.

For the production design, of more than 100, we  plan to use  a smaller formfactor,  based on 400MHz ARM computer , as in image below.  This will have a spec similar to above camera , with a microphone, and Real Time Clock.







Thermal image of Marks and Spencer's bakery, spot the freshest bread :-)
Taken with FLIR camera, not yet in SenseCam. 




For more details of the LifeCloud sensecam lifelogging camera please  contact Lyndsay Williams sensecam@gmail.com or +44 (0) 7970 101578. Cambridge, UK


"Sensecam"  name copyright Lyndsay Williams.
All intellectual property  and designs in this blog owned by Lyndsay Williams, unless mentioned otherwise.



Saturday, 10 September 2016

Apple Airpods - a new health sensing platform?




Inline images 2
Apple Airpods have just been announced, many customers are questioning the high expense, removal of the traditional 3.5mm headphone socket and how easily Airpods might be to get lost.

Here is the Apple video explaining the Airpods.

These could be much more than music headphones however. 

The Airpods contain an infrared sensor, accelerometer, microphone  and possibly  other sensors, this could provide a brand new wearable hardware platform for health and medical sensing. For example core body temperature, maybe heart rate detection. The accelerometer could be used for head position sensing e.g. head nods "yes" or shaking "no". This could be very useful for some applications.
Head accelerometers have also been used to measure concussion in sport, e.g. football.
The Airpods will use bidirectional  digital communication, so could not use the old analogue 3.5mm  socket.  At least with Bluetooth, if you drop the Airpods, you can find them if location within a few metres, the iPhone will show nearby Bluetooth devices.


Image below from Apple of inside an Airpod.


Inline images 3
 We can see the infrared sensor, the dark cylinder in image, possibly a  thermopile which allows none contact temperature sensing, e.g. of the human body in seconds.  Here an example of a thermopile from Texas Instruments. It is the small reflective semiconductor, U5, size 1.8mm x 1.8mm.







Below shows a thermal image of an ear, revealing the core body temperature, a healthy 38C. The colder blue is the outer ear. This image taken with an array of 80 x 60 thermal sensors.



Inline images 2



  Passif designed the custom wireless semiconductor. More here about Passif technology.  An interesting patent from Passif in this Fast Company article describes an energy harvesting system from radio waves.


  
Inline images 1
 
Apple own the earliest patents on a mobile phone with infra red  human proximity detection,  and accelerometer control.  I invented and built  a handheld computer  design in 1997, the first phone that used these sensors. The patent was purchased by Apple in 2011 as here, patent for Apple.