Eric Manser and Will Scott from IBM Research, presentation on "Cognitive Insights Drive Self-driving Accessibility" as part of the Cognitive Systems Institute Speaker Series
2. Important Factors
• Over one billion people, or 15% of the world's
population, experience some form of disability (World
Health Organization)
• According to the United Nations, the number of people
aged 60 years or older is projected to grow by 56
percent worldwide by 2030.
• 1.6 billion people or 23% of the world’s population is
over age 50
• The world is experiencing an explosion in connected
devices and cognitive computing capabilities, with over
6.4 billion connected devices and interfaces worldwide
• Estimated that 10 million self-driving cars will be on the
road by 2020.
People with
Disabilities
New Technologies
Aging Population
Self-Driving
Vehicles
3. About the Project
IBM Research and Watson IoT have announced a new
collaboration with Local Motors and the Consumer Technology
Association to create the world's most accessible, self-driving
vehicle.
• This partnership will leverage the power of IBM Watson, the
Internet of Things, and new innovations in accessibility
technology.
• We will be launching a series of workshops with talented people
of all ages, backgrounds and abilities, including P-TECH
schools, High School STEM, AARP and universities including
MIT and Princeton
• Creating the most accessible vehicle will:
• Allow older adults to age in place and remain independent
and self-sufficient in their communities
• Help those with cognitive, vision, memory or physical
challenges to leverage transportation, improving their
independence and quality of life
• Provide enhanced access to work opportunities and
community services to people of all ages and abilities
• Ensure a successful transition to driverless cars for all
5. P-TECH at Carver Workshop
Location: Local Motors National Harbor MD
Date: February 9, 2017
Audience: PTECH Carver Students
Brainstorm Challenge - High Impact Area
Recommendations
AARP Hatchery Workshop
Location: AARP Hatchery, Washington DC
Date: February 10, 2017
Audience: Older adults (60+)
Recommendations
• How would you know what to do in an emergency or dangerous
situation (can't hear or see)
• How to safely find and get to an Olli stop
• How do I get on and off Olli
• How do I know where to sit
• How do I know where I am at and when to get off; missing stops
• How do I interact with other passengers (can't hear, see, speak)
• Make Olli On Demand - Olli should come to the door, pickup
at exact time and location via easy to use call application
• Make Olli Safe – should be able to avoid danger/accidents,
ensure safety at stops and onboard; give emergency
instructions and call to nearest law enforcement and first
responders based upon location
• Make Olli Personal - should recognize/scan riders before
entering vehicle, know their preferences and selected
profile information
• Make Olli comfortable - easy to get on/off, ability to
accommodate or stow equipment (cane, wheelchair,
umbrella, etc..), seating that adjusts for bone/joint issues
• Make Olli easy to understand - Information delivery should
be clear and delivered in multiple formats - visual and
auditory
• Make Olli affordable to use
6. Front Porch Workshops
Location: Retirement and assisted living centers in San
Diego, Carlsbad and La Jolla, CA
Date: February 27-28, 2017
Audience: Older Adults (60+)
Recommendations
Visual Impairment Workshop
Location: IBM Research, Cambridge MA
Date: March 7, 2017
Audience: Advocates for Visually Impaired and
Transportation (Mass Association for the Blind,
Perkins School, Umass, Carroll School, MBTA)
Recommendations
• Alerts/Notifications - Need multiple interfaces – audio, text, haptic
• Develop bluetooth connection for headset coupled with a
smartphone/pad app - can provide personalized instructions and
experience
• Provide an Alexa-like service or app to call for or schedule an Olli
• Need secure way to access and utilize vehicle - fob key, biometric
or visual recognition
• Improve access on/off for wheelchair users and others with
assistive devices
• Low riser for entrance/exit
• Provide storage for walker, canes, packages
• Develop a seat that can raise or lower automatically to help with moving
between sitting and standing positions; provide options for individual
seats
• Implement grab bars with certain seats
• Accessibility features should be discreet - ramps should raise/lower
once user who needs them is detected.
• Have Olli recognize rider and implement personalized
profile option
• Have a personalized narration delivered over smart
device/headphones; deliver ability to repeat info
• Voice enabled navigation for entrance/exit and seat location
• Provide ability to secure wheelchairs and other assistive
devices
• Some vehicles have built in clamps/restraints - others
secure via straps; no standards in place from wheelchair
manufacturers, depends on device and vehicle
• Provide a pre-made wheelchair for Olli
• Provide moveable/retractable seats to accommodate
wheelchairs
• Provide redundancy for wheelchair accessible entrance and
exit
• Accommodate service animals
7. HLAA and Henry Claypool Discussions
(Hearing and Mobility Impairment)
Date: March 29, 2017
Recommendations:
Hackathons
MIT ATHACK, Cambridge, MA
Date: March 4, 2017
Challenge: Create a system to guide a visually impaired
rider to an open seat on a public vehicle
Solution:
• Overhead camera analyzes interior images to identify
open and occupied seats
• Rider pairs smartphone to vehicle
• Camera communicates with smartphone app
• App opens in window the size of the screen,
simplifying engagement
• App communicates via Bluetooth audio navigation
guidance to rider
• As a baseline, all information should be available via text –
app, exterior of vehicle and interior
• Sign language translation highly desirable – sign to Olli, and
receive sign language back
• Voice activated systems need to integrate with Tcoil
loops/assistive devices
• Integration with Bluetooth devices desirable
• Integrate Haptic (touch) technology for alerts
• Provide for ambient noise reduction - phone/listening cone
at each seat; have info directed to telecoil/hearing aid
• All information – route, destination, updates, safety
instructions, alerts – should be provided via multiple
interfaces – visual, text, auditory with Tcoil loop assist, sign,
haptic
• Need to provide automated lock down for wheelchairs
• Need to provide space to turn wheelchair around in vehicle
8. Hackathons
HackPrinceton
Date: March 31, 2017
Future Workshops & Hackathons
Workshops
• City of Columbus/OSU – Mobility, TBD
• City of Las Vegas, TBD
• City of Santa Clara/Palo Alto VA
Hackathons
• TOM-NYC (Tentative), Brooklyn, NY
Events
• AAPD May 17, Local Motors National Harbor
• M-Enabling Summit, June 13-14, Washington DC
• Watson Developer Conferences, 9 locations worldwide
• Q2-Q4
Challenge: Win the Best Use of Watson IoT Challenge by
creating an innovative Accessibility solution using
#AccessibleOlli.
Results:
• 12 teams built with Watson. 67 teams overall submitted
projects.
• 500 developers engaged, 50 enabled with Watson IoT
platform
10. Potential Solution Ideas
• Uber-like app for Olli, voice and text interfaces
• Personalization is a key component - Olli should be able to
recognize your face or your fingerprint, and know about
what route you usually take, what your schedule is,
interaction preference, language preference, who to contact
in an emergency
• All interactions on Olli should be multi-modal – voice, text,
touch
• Have a personal "Olli in your ear" that guides you on your
trip - have Olli describe what you are passing, give you
reminders and suggestions, and tell you when you have
arrived at your stop
• Use voice, text and haptic or other technologies to guide
people on the vehicle and find open seats; Seats should
vibrate to signal a stop
• Olli should have a seat that automatically folds away to
accommodate a wheelchair user and a place to store
walker or other assistive mobility devices. Wheelchairs and
other assistive devices should automatically be secured.
• Have Olli detect emergency situations either in or outside
the vehicle, and call the nearest responders (EMT, police,
fire) for help. All emergency instructions should be given
Focus on riders with these challenges:
• Visual impairments, from low vision to complete blindness
• Cognitive issues, ranging from dyslexia to memory loss
• Hearing impairments including deafness
• Mobility issues requiring assistive devices such as
wheelchairs, canes and walkers
For each category, use cases will cover
• Arranging for a ride on Olli
• Navigating to the vehicle
• Navigating onto the vehicle
• Securing a seat/location on the vehicle
• Experiencing the ride and being alerted to destination
• Arriving at destination and exiting the vehicle
11. Arranging for a ride on Olli Navigating to Olli Navigating onto Olli
& Securing seat
Experiencing ride and
destination alert
Arriving at destination
and exiting vehicle
Use Case Template
Cognitive “Middleware” Across All Use cases
Multi-Modal Interface
• Voice
• Text
• Haptic
Personalization
• Profile
• Language
• Interface
• Alerts
Security
• App or biometric access
Emergency contacts and
process
Visual
Hearing
Mobility
Cognitive
Directions & Guidance
• Location Support
• Curb cut-outs
• Ramps
• Elevators
Integration with App or Device
12. Arranging for a ride on Olli Navigating to Olli Navigating onto Olli & Securing seat Experiencing ride and
destination alert
Arriving at destination
and exiting vehicle
Visual Use Case
Multi-Modal InterfacePersonalization Security Directions &Guidance
“Uber” or “Alexa” –like app,
voice activated
• Establish secure profile with
preferences
• Call/arrange for Olli pickup
• Get updates on
arrival/timetable
• Get details on location
Using voice-activated
smartphone get:
• Updates on location
• Arrival time
• Any route issues/delays
• Navigation guidance to
pickup site
Smartphone app should sync with Olli before boarding
and communicate/confirm via Bluetooth w/headset:
• Onboard info
• Identification of rider
• Destination and preferred route
• Preferences (language, interface)
• Emergency contact
• Navigation
• Using VR and interior camera, identify
open seats
• Communicate open seat location to rider
via app (voice)
• Communicate location of storage area
(canes, packages, etc..
Smartphone app to provide Olli in
your ear/app” – voice interaction
via Bluetooth integration
• Pre-notification alert
• Arrival alert at destination.
• If desired, route narration
• Emergency situations
• All instructions given via
voice
• Emergency personnel/first
responders alerted that
riders with disabilities are on
board
Delivered via
Smartphone via
voice
• Destination
announcement
• Reminder and
retrieval of stored
items
• Rider advised of
way out and other
accessibility
features at
destination
Cognitive middleware integration with voice activated Smartphone app
14. Existing Watson Integration with OlliPotential New Technology Areas
• Visual Recognition
• If trained on wheelchair/walker, automatically lower
ramp, fold seat
• Detecting someone is at a stop and/or has ordered
Olli
• Rider recognition
• Open/Occupied seats
• Emergency situations
• Personalization
• Rider profile – created by caregiver or individual
• Interface and Language choice
• Destination options
• Reminders
• Notification is rider gets off/on at non-profile locations
• Haptic
• For blind/low vision aid with navigation on/off and with
finding seat
• Haptic feedback to notify rider they are at destination
• Bluetooth integration
• Braille reader integration
15. Watson Analytics and Machine Learning
• There are a number of Watson services in use with Olli today –
the next step is to incorporate Watson Analytics and domain
specific Machine Learning algorithms
• Watson Analytics is a cloud-based, smart data discovery service
• Guides data exploration
• Automates predictive analytics
• Machine learning is a form of artificial intelligence (AI)
• Provides computers with the ability to learn without being
explicitly programmed
• Computer programs that can change when exposed to
new data
16. Data Driven Insights
• Time spent in transit can be leveraged to yield many data driven
insights
• Big data and analytics, coupled with cognitive computing
approaches can be used
• Analysis can be performed to asses general health and well-
being of passengers over time, using non-intrusive sensors to
monitor:
• Weight – via seat mounted sensors
• Heart rate – via infrared cameras
• Facial expressions (for emotional well-being) – via
machine learning and visual recognition
• Gait - via machine learning and visual recognition
• Cognitive decline – via onboard digital “games” to access
mental well-being
• This data can be fed to a larger, holistic cognitive solution…
17. Considering a Holistic Cognitive Solution
• A cognitive Knowledge Aggregation solution is
embodied by a contextual data fusion engine,
that centralizes IoT and System of
Record/Engagement data across multiple
channels
• “Snapshots” of passenger data captured while
riding Olli can be applied to this larger, holistic
cognitive solution to monitor passenger health
and well-being
• Over time, Olli-based data, along with data
captured from other channels (home, wearables,
social, etc.) can be used to predict inflection
points in well-being