Machine intelligence to free human intelligence: How automation helps you win

  • View
    283

  • Download
    2

Embed Size (px)

Transcript

  1. 1. Machine intelligence to free human intelligence: How automation helps you win Roger Chen @rgrchen roger@oatv.com OReilly Solid 2015 June 25, 2015
  2. 2. Theres been a lot of excitement about robotics & AI recently the development of full artificial intelligence could spell the end of the human race. I think its a distraction from the conversation aboutserious issues, "With artificial intelligence, we are summoning the demon"
  3. 3. Contrary to what some would have you believe
  4. 4. robot AIs are not going to end humanity anytime soon
  5. 5. So dont be awkward about it
  6. 6. And get to know the machines that can help you
  7. 7. Whats going on? How did this happen? Whats new this time around? Rules of engagement for automation
  8. 8. So whats all the fuss about? Robots arent new
  9. 9. but these ones are
  10. 10. Big companies are innovating too Source: BCG
  11. 11. Robotics is accelerating 0 50 100 150 200 250 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Industrialrobots(000s) Global industrial robot sales
  12. 12. Globally South Korea Japan Germany US China Industrial robots operating by region Industrial robots per 10,000 manufacturing employees 30 437 323 282 152 307K 237K 182K 176K 175K
  13. 13. How did this happen?
  14. 14. Smartphones
  15. 15. How did this happen? Smartphones Sensors
  16. 16. Novel, cheaper sensors
  17. 17. How did this happen? Smartphones Sensors Actuators
  18. 18. Actuator & servo innovation Compliance (safety) Cost Performance Series elastic actuator (1993) Further actuator innovation (1993 ) First low-cost robotics startups (Baxter, 2008) New startups pushing envelope (Modbot, 2015)
  19. 19. How did this happen? Smartphones Sensors Actuators Software
  20. 20. Open source software: Robot Operating System (ROS) Collaboration Standardization Lowering technical barriers From 20 man-years to 18 months
  21. 21. ROS proliferating in technical world 0 200 400 600 800 1000 1200 1400 1600 1800 Citations Total citations over time
  22. 22. 0 10000 20000 30000 40000 50000 60000 Jan-07 N ov-07 Sep-08 Jul-09 M ay-10 M ar-11 Jan-12 N ov-12 Sep-13 Jul-14 IPs Unique IPs downloading packages Switchyard (Feb 2007) ROS at Willow Garage (Nov 2007) ROS at ICRA (May 2009) ROS 0.4 release (Feb 2009) ROS being put to use
  23. 23. ROS global
  24. 24. How did this happen? Smartphones Sensors Actuators Software Market
  25. 25. Consumers Online retail sales $262B (US, 2012) 10% CAGR Logistics and transportation Total spending $1.33 trillion (US, 2013) 8.5% of US GDP Customers (brands) Want more for less to satisfy consumer demand Competitors Are undercutting one another on price Faster Xiaomi $12B sales 135% Cheaper Personalized Intense market forces Workforce 600K unfilled mfg jobs 25% turnover in MH&L
  26. 26. Whats new this time around?
  27. 27. Whats new this time around? Its not just hardware
  28. 28. Rise of data and machine learning Machine learning algorithms (deep learning) Explosion of training data Hardware (GPUs) Interest and resources
  29. 29. Deep learning
  30. 30. Explosion in training data
  31. 31. Perception automation
  32. 32. GPUs
  33. 33. Whats new this time around? Connectivity Connect, measure, control
  34. 34. DevOps in the physical world I I I I 0 I I 0 I 0 I 0I 0 0 I 0 I 0 I I I I I 0 I 0 0 II 00 I I 0 I I 0 I 0 I I I I I0 0 I 0 I 0 0 0 0 I I I 0 I I 0 I 0 I I 0 I 0 I I 0I 0 I 00 I I I 0 0 I 0 I 0 I 0 0 I I I 0 0 0 0 I I I 0 0I I I I 0 I 0 0 I I I 0 I 0 I0 Machine learning I 0 I 0 I 0 I 0 0 I I I I I 0 0 I I 0 I 0 I 0 I 0 0 I I I I I 0 0 I I 0 I 0 I 0 I 0 0 I I I I I 0 0 I I 0 0 I 0 I I I I I 0 0 I I 0 I 0 I 0 I 0 0 I I I I I 0 0 I I 0 I 0 I 0 I 0 0 I I I I I 0 0 I I 0 I 0 I 0 I 0 0 I I I I I 0 0 I I I 0 I 0 I 0 0 I I I I I 0 I I 0 I 0 I 0 I 0 0 I I I I I 0 0 I I 0 I 0 I 0 I 0 0 I I I I I 0 0 I I 0 I 0 I 0 0 I I I I I 0 0 I I 0 I 0 I 0 I 0 0 I I I I I 0 I 0 0 I I 0 0 I I I I 0 I 0 I I 0 0 I 0 0 I 0 I 0 I 0 I I I 0 I 0 0 I III I 0 I I 0 0 0 I I I I 0 I0 0 I I 0 I 0 I 0 0 I I 0 0 I I 0 0 I I I I 0 0 I I I I 0 0 0 I 0 0 I I 0 I 0 I 0 0 0 I I I I I 0 0 I 0 I I II I I 0 I 0 I 0 0 I I 0 I I 0 II 0 0 I II I I 0 I I I 0 I 0 0 I 0 I I I 0 I 0I I 0 0 II I 0 I I 0 0 I II I I 0 0 I I I I I 0 0 I I I I I 0 0 I I I I I 0 0 I I 0 I 0 0 I 0 0 I 0 0 0 Deploy
  35. 35. Whats new this time around? Flexible automation
  36. 36. Flexible No days off!Keep working! Pick items for shipping Pack manufactured goods Clean Rogers car Do some research Transport some boxes
  37. 37. Why flexible automation matters ROI ~ (ResultsA + ResultsB + ResultsC)/($A + $B + $C) Project B Project A Achieve ROI across multiple deployments Project C Machines work capacity
  38. 38. Fixed to flexible to programmable Production volume Low Medium High Variety High Medium Low Programmable Flexible Fixed Too hard Poor ROI
  39. 39. Whats new this time around? Cost and compliance
  40. 40. Lower cost and compliance open up new applications $25K $34K
  41. 41. Fetch Robotics Logistics Warehousing Material handling Transportation Manufacturing
  42. 42. Plethora Labs Manufacturing
  43. 43. Emerald Cloud Lab Life science lab robotics
  44. 44. Riffyn R&D data automation DesignMeasureImprove Make it reproducibleStack it Build protocols like Legos Automate data acquisi=on Automate error checking & root cause analysis Share and version like Git Collaborate Riffyn
  45. 45. Rules of engagement for automation
  46. 46. It has to work 2+2=3?!
  47. 47. ROI If it doesnt make dollars, it doesnt make sense Cost savings Productivity Quality Turnaround Scalability Reproducibility New opportunities
  48. 48. Machines for problem reduction, not the entire solution Human + machine Machine Perfect memory Complex statistical analysis Repeatability Endurance Human Intuition Creativity Inference Abstraction But this matters!
  49. 49. Human + machine + process Weak human + machine + better process was superior to a strong computer alone and, more remarkably, superior to a strong human + machine + inferior process. -Garry Kasparov, Chess Grandmaster The Chess Master and the Computer Everyone should read this!
  50. 50. Use machines, intelligently Machines work capacity ROI ~ Work/$ ROI ~ Results/$ Garbage work Useful work Garbage in, garbage out Process matters work smart, not just hard What problem are you really solving? If you struggle to fill capacity, automation may not be right for you
  51. 51. Challenges Technology Perception, deployment, programming Implementation Safety, security, cultural acceptance Society Impact on labor and economy Machines are reinventing work
  52. 52. What does the future of work look like?
  53. 53. Theres plenty of room at the bottom.
  54. 54. Theres plenty of room at the bottom. Theres also plenty of room at the top.
  55. 55. Thanks! Roger Chen @rgrchen roger@oatv.com