Booking open Available Pune Call Girls Pargaon 6297143586 Call Hot Indian Gi...
A.Levenchuk -- visuomotor learning in cyber-phisical systems
1. Cyber-physical systems architecture breakthrough:
learning of visuomotor coordination
Moscow
9-dec- 2015
107th meeting of INCOSE Russian chapter
2. Cyber-physical systems
• Cyber-Physical Systems or “smart” systems are co-
engineered interacting networks of physical and
computational components
• CPS include:
- Internet of Things (IoT)
- Industrial Internet
- Smart Cities
- Smart Grid
- "Smart" Anything (e.g., Cars, Buildings, Homes,
Manufacturing, Hospitals, Appliances)
• NIST CPS Public Working Group --
http://www.nist.gov/cps/index.cfm
• NSF -- http://cps-vo.org/ 2
3. Draft Framework for Cyber-Physical Systems
3
http://www.nist.gov/el/nist-releases-draft-framework-cyber-physical-systems-developers.cfm
This is all about
systems engineering!
4. From «smart» to «intelligent»
How CPS perform it Decision?
Sensors
Consoles
Actuators
Monitors
http://www.nist.gov/el/nist-releases-draft-framework-cyber-physical-systems-developers.cfm 4
5. Where is that «intelligence»?
Cyber-physical device
Software
Interfaces and
communications
Cognitive
processing
Hardware
Sensors Mechanics Actuators
5
6. Knowledge engineering
• Decision is carefully programmed
(manually).
• Example: robot-«butterfly»,
https://youtu.be/kyvW5sOcZHU, https://youtu.be/V30e77x8BQA
• Every type of movement should be programmed anew
• Non-adaptable to changes of environment and device
• The best science available up today!
• Perfect, if CPS perform only one or two movements. Not for
robots, definitely!
6
7. Goal: CPS capuchin-like
• Jurgen Schmithuber (July 2015): In order to pick a fruit at
the top of a tree, Capuchin monkey plans a sequence of
sub-goals (e.g., walk to the tree, climb the tree, grab the
fruit, …) effortlessly. We will have machines with animal-
level intelligence in 10 years.
https://sites.google.com/site/deepernn/home/blog/briefsummaryofthepaneldiscussionatdlworkshopicml2015
• Needs planning
• Needs great visuomotor coordination!
• Impossible to program manually up to date.
7
8. New in «decisions»:
learn to decide!
Machine learning and reasoning:
• Symbolic (by induction)
• Evolutionary (by genetic programming)
• Bayes (by probability assesement)
• By analogy
• Connectivist (deep learning, artificial neuron nets)
8
The Master Algorithm: combine ‘em all!
http://www.amazon.com/dp/0465065708/
14. DeepDriving
• train a deep Convolutional Neural Network (CNN) using 12 hours of
human driving in a video game
• show that our model can work well to drive a car in a very diverse set of
virtual environments
• train another CNN for car distance estimation on the KITTI dataset,
results show that the direct perception approach can generalize well to
real driving images
• Open sourced
• Autopilot Driving is not a miracle now: Tesla X, Google car,
AVRORA/KAMAZ, Volvo trucks, and counting
14
http://deepdriving.cs.princeton.edu/
15. Cortical sensory
homunculus
• Body is an easy part.
• Manipulation is difficult!
• Non-prehensile manipulation is included.
15
https://en.wikipedia.org/wiki/Cortical_homunculus
16. Visuomotor examples
• End-to-End Training of Deep Visuomotor Policies,
http://arxiv.org/abs/1504.00702
• Supersizing Self-supervision: Learning to Grasp
from 50K Tries and 700 Robot Hours,
http://arxiv.org/abs/1509.06825, https://youtu.be/oSqHc0nLkm8
16
17. Self-learning robots
• Fanuc: $7.5mln for 6%
in Preferred Networks
• ABB invested up to
$10mln in Vicarious
http://www.bloomberg.com/news/articles/2015-12-03/zero-to-
expert-in-eight-hours-these-robots-can-learn-for-themselves
Osaro -- http://www.osaro.com/, learning
in environment, including cooperation
with humans,
http://www.technologyreview.com/news/543956/a-supercharged-system-
to-teach-robots-new-tricks-in-little-time/
17
18. Non-prehensile example
• Deep Spatial Autoencoders for Visuomotor
Learning,
http://rll.berkeley.edu/dsae/ (video)
18
20. Visuomotor hardware
20
Quantum computer: waiting, but promising – currently 100mln times faster
than classical desktop (http://googleresearch.blogspot.com/2015/12/when-can-quantum-annealing-
win.html).
21. Visuomotor sensors
• Computational multi-lens optics (near future)
• Solid state LIDARs for driving (below $100 in five
years, now $1000) -- http://www.quanergy.com/,
http://velodynelidar.com/, includes processor and
neural software
• But Elon Musk tell that lidar not needed, only
optical cameras and radar
21
http://blog.lidarnews.com/lidar-for-self-driving-cars/
22. Voice interface for visuomotor goal settings
• Deep learning is a leading technology for a voice recognition
• Voice command interface is not a problem today
• General intelligence of a CPS is a problem!
Company Name of Personal assistant
Google Google
Apple Siri
Microsoft Cortana
Facebook M
Amazon Alexa
[Cyber-physical system
vendor]
???????????
22