Autonomous systems are like teenagers. The decision to trust one to complete a task without strict supervision depends on the individual and the task. Performance and trust are variables on spectra. As with teens, some autonomous systems will be ready before we are ready to trust them, and some will take a little longer.
As we get comfortable with delegating routine domestic tasks to home robots and prepare for a world with self-driving cars and beyond, it is important to understand the opportunities and limits for autonomous systems. Participants in this webinar will learn about the technologies that enable autonomous systems, and how to critically assess design constraints for independent and collaborating autonomous solutions.
Smart Data Webinar: The Road to Autonomous Applications
1. The Road to Autonomous Applications
Adrian Bowles, PhD
Founder, STORM Insights, Inc.
Lead Analyst, AI, Aragon Research
info@storminsights.com
DECEMBER 14, 2017
2. • Overview and Opportunities
• Enabling Technologies
• Autonomous Application Categories
• A Matter of Trust
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
AGENDA
3. ROBOCOP IS COOL, BUT…ROOMBA IS REAL
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
4. Autonomous: The ability to make one’s own decisions.
Automatic: A system that responds to environmental input with pre-programmed responses.
Semi-autonomous: A system capable of making [some] decisions based on
context, and relying on human intervention or override for others.
A single system may have multiple modes.
AUTONOMOUS VS AUTOMATIC
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
6. Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
Autonomous Intelligent
AUTONOMY VS INTELLIGENCE
Abstract
Generalize
Learn
Understand
ReasonIndependent
7. Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
Listen
Look
Touch
Smell
Taste
Your System
Learn
Model
Reason
Understand
INTERACTING WITH THE OUTSIDE WORLD
13. BigDog is a dynamically stable
quadruped robot created in 2005 by
Boston Dynamics with Foster-Miller, the
NASA Jet Propulsion Laboratory, and the
Harvard University Concord Field
Station.[1] It was funded by DARPA, but
the project was shelved after the BigDog
was deemed too loud for combat.[2]
BigDog. (2016, August 19). In Wikipedia, The Free Encyclopedia. Retrieved 19:15, September 7, 2016, from https://en.wikipedia.org/w/index.php?title=BigDog&oldid=735286400
MQ-9 Reaper
Unmanned Areal Vehicle (UAV/Drone)
SOMETIMES, A LITTLE AUTONOMY GOES A LONG WAY.
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
14. WHEN DO WE NEED HUMANS IN THE LOOP?
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
15. SAE AUTONOMOUS VEHICLE CLASSIFICATION
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
17. Google Patents Emergency Vehicle
Detection for Autonomous Cars
9/6/2016
INTEROPERABILITY WITH THE EXISTING INFRASTRUCTURE IS CRITICAL
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
18. Autonomous Systems Artificial General Intelligence
Machine
Learning
Cognitive ComputingThe Internet of Things
Understand
Reason
Learn
Domain-independent
problem solving
Connectivity
Control
AUTONOMOUS SYSTEMS IN CONTEXT
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
Enabling Technologies
IoT……Sensors,
Intelligence at the Edge…
Vision Technologies…
LIDAR
19. Supervised Unsupervised
Deep
General
Reinforcement
Learning by example,
using training data. Strategies based
on performance
feedback.
Discovers patterns based
on experience with data.
Biologically-inspired ML approach.
Leverages simple processing units - analogous to neurosynaptic elements -
organized in layers that collaborate to solve complex problems.
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
MACHINE LEARNING FOR AUTONOMOUS APPLICATIONS
20. Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
CONTRASTING AI APPROACHES USED FOR INTELLIGENT AUTONOMOUS SYSTEMS
Knowledge-Centric Data-Centric/
Deep Learning
Representation Learning
Use ML to discover the representation
Lots of Up-Front Effort
Developing the Algorithms
or Rules
Should have
Complete Transparency
Identify the Categories
Let the Data Drive the Process
Can Become a Black Box
ATTRIBUTES
APPROACH
Use ML to discover the mappingUse experts to create the
representation and mapping
21. Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
HOW TO DELEGATE AUTONOMY
ALGORITHMS
&
RULES
DATA
22. AutoSys
Decision-making…
Does it plan?
Generative planning?
Use feedback?
Coordinate?
Can it:
move?
in the air?
on land?
on/under water?
see?
hear?
smell?
taste?
feel?
learn?
CLASSIFYING AUTONOMOUS SYSTEMS
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
Segmentaton
23. Train
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
A MATTER OF TRUST
Verify
Validate
Delegate
Evaluate
TRUST
DecisionsPreparation
24. Copyright (c) 2017 by STORM Insights Inc. All Rights reserved.
PROGRESS?
25. Alex Coppel / Herald Sun/Getty Images
Fernando Alonso Walks Away from a 46G Impact
Will F1 become fully autonomous?
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
FROM INFORM/ADVISE TO AUTOMATE
26. DESIGN CONSTRAINTS - DECISIONS UNDER DURESS
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
27. Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
DESIGN DECISIONS/CONSTRAINTS
Independent vs Collaborative
Execution Constraints
Augment vs Automate
29. •A robot may not injure a human being or, through inaction,
allow a human being to come to harm.
•A robot must obey orders given it by human beings
except where such orders would conflict with the First Law.
•A robot must protect its own existence
as long as such protection does not conflict with the First or Second Law.
“Handbook of Robotics, 56th Edition, 2058AD”
in
Runaround, a 1942 short story
ISAAC ASIMOV - THREE LAWS OF ROBOTICS
Copyright (c) 2017 by STORM Insights Inc. All Rights Reserved.
30. adrian@storminsights.com
Twitter @ajbowles
Skype ajbowles
If you would like to connect on LinkedIn, please
let me know that you that you registered for the
Smart Data webinar series.
KEEP IN TOUCH
2018 SmartData Webinar Dates & Topics
January 11 AI At The Edge:
Pushing Intelligence to Fog Computing Nodes
February 8 A Pragmatic AI Maturity Model:
Choosing the Right AI Technologies Based on
Application Requirements and User Attributes
March 8 Machine Learning Update:
An Overview of Technology Maturity
and Product Vendors