Socially Assistive Robotics Grace Chandler Socially assistive - - PowerPoint PPT Presentation

Socially Assistive Robotics

Grace Chandler

Socially assistive robotics (SAR):

An intersection between assistive robotics and socially intelligent robotics, often for a therapeutically beneficial role

TEACHING: MIT’s Tenga- A second language learning companion

ELDER CARE/MENTAL HEALTH: AIST’s PARO, a pet-like plush robot- Improves socialization and motivation

COMPANIONSHIP: PaPeRo (Partner-type Personal Robot) is a development platform for a robot companion.

AUTISM: RoboKind’s Milo- an anthropomorphic robot that teaches social behaviors

Vocabulary and Definitions

• RCT : randomized controlled test
• CBT : cognitive behavioral therapy
• ASD : autism spectrum disorder
• Cronbach's alpha (): a measure of internal consistency,

how closely related a set of items are as a group.

Paper 1:

Integrating socially assistive robotics into mental healthcare interventions:

Applications and recommendations for expanded use

Rabbitt, Kazdin, Scassellati

Motivation

• Mental illness is a pervasive issue

○ ¼ Americans meet criteria in a given year ○ ½ Americans meet criteria at some point in their lives

• The majority of the mentally ill receive any form of

treatment ○ Even fewer receive evidence based treatment ○ We are not meeting treatment needs

Socially Assistive Robotics as a Solution

• The one-to-one therapy model that is common requires

unmanageable resource requirements

• Socially assistive robots can lighten the load on the mental

health care system ○ Provide care where it might be otherwise unavailable ○ Provide around the clock care when clinicians are unavailable ○ Encourage and monitor adherence to treatment

Goal of the Paper

• Use evidence gathered from other socially assistive

projects to prove the viability of application to mental health care

• Detail the success of other projects
• Define the working space and demographic of potential

robots

• Address obstacles of implementation

Types of Socially Assistive Robots

• Animal-Like
• Anthropomorphic
• Caricature
• Machine-Like

Applying Existing Proof to Mental Health

• Bandit: SAR as a motivation tool resulted in more positive

interactions in an exercise program for the elderly

• Autom: SAR as a tool to encourage program adherence

was demonstrated with a weight loss robot

SAR Roles:

• Companion
• Coach or

Instructor

• Therapeutic

play partner

Companion Role

• Substitution for therapy animals
• Can reduce loneliness and social isolation
• Aibo was shown to have the same benefits as a therapy

dog

It is important to note:

Success != Performing better than current method Success == Producing at least the same benefits as current method

Therapeutic Play Partner Role

Work alongside human caregivers to: ○ Increase engagement ■ Keepon, Pleo ○ Offer opportunities for social interaction ■ I.e: modeling and enforcing social cues

Coach or Instructor Role

Provide:

• Corrective feedback
• Support
• Encouragement
• Task modeling

Integrating SAR Into Existing Programs

• Can deem SAR effective faster because the existing

program has been already deemed effective

• Probo was integrated into existing Social Stories program

○ Compared to human therapist ○ Behaviors targeted by program improved significantly more with Probo

What Needs to Be Done

• Expand breadth of clinical application to generate data

○ The mental health crisis will get worse, we need to catch up

• SAR projects need to be proven to be evidence-based

○ Single-Case designs can speed this up ○ Implementing SAR into existing treatments also helps

• Roboticists and clinicians need to collaborate

Conclusion

The magnitude of the mental health crisis in America cannot be handled by existing resources alone. SAR provides an

• pportunity to supplement existing resources, rather than

attempting to replace human clinicians and therapists. While there are hurdles to address, it is critical that the Psychology and Robotics community put effort into this field.

Paper 1 Discussion

• What did we like about this paper?
• What didn’t we like?
• How do you feel the paper addressed the topic of mental

health?

• Are robots an adequate supplement for therapy treatment?
• Were all of the cited examples relevant?

Paper 2:

Brian 2.1:

A Socially Assistive Robot for the Elderly and Cognitively Impaired

McColl, Louie, Nejat

Problem

• The elderly population is continuously growing
• The elderly are at risk for developing cognitive decline

○ Age related memory loss ○ Dementia ○ Alzheimer’s

• 115 million will have age related memory loss by 2050

Motivation

• Cognitive decline leads to:

○ Difficulties performing self-care ○ Inability to live independently

• Cognitive training interventions have been shown to

positively affect cognitive functioning in older adults

Goal

• Advance knowledge of the effects of cognitive and social

interventions for elderly populations suffering from cognitive decline.

• Design a human-like platform to test these interventions

Approach

Brian 2.1 : ○ Anthropomorphic robot ○ Determines appropriate behavior based on the user’s state and current activity state ○ Delivers one-on-one interaction for eating and entertainment

Robot Design

• Waist-up human model

○ 3 DoF neck ○ 2 DoF waist ○ 4 DoF arms (2)

• Waist-up human model

○ 3 DoF neck ○ 2 DoF waist ○ 4 DoF arms (2)

Eating Assistance Task

• Weighted tray which tracks

○ What is currently being used/last time it was used ○ How much of certain foods has been eaten

• Utensil tracking and face tracking Emotions:

○ Encouragement → happy ○ Orientation → neutral ○ Long term distraction → sad

Card Game Task

• Leads users through memory card game
• Acts as coach
• Provides instructions, hints and reinforcement
• Emotions:

○ Instruction → neutral ○ Encouragement → neutral ○ Long term distraction → sad ○ Celebration → happy

Monitoring Attention and Investment

• Monitoring utensil use
• Keeping track of food weight
• Body Language

○ Tracking head position ○ Determining trunk position ○ Trunk and head position combined were used to determine how accessible the user was

User Study

• Placed in public space at long-term care facility for two

days

• Robot introduced itself and offered to play a card game

○ Only explained meal capabilities

• Members of research team were present

Metrics for Assessment

• Duration of interaction
• Engagement in interaction

○ Frequency and type of participant interaction

• Compliance and cooperation
• Acceptance and attitudes toward robot

Results

• Majority engaged and complied
• 82% smiled or laughed when Brian displayed happy

emotions

• 57% were reengaged by Brian’s sad state and felt

empathetic

• 43% were reengaged by bringing focus back to the robot

and the activity

Conclusion

• Elderly residents were accepting of the robot
• Residents voluntarily used the robot
• Residents were engaged by the robot
• Brian’s emotions were well-received
• There was no significant difference in use based on

gender

Paper 2 Discussion

• What did we like about this paper?
• What didn’t we like?
• How do you feel about their results? Were they believable?

○ Did they demonstrate the full capabilities of the system?

• What are your thoughts on implementing this robot on a

larger scale?

• Was the approach for encouraging behaviors appropriate?
• How do we feel about their discussion of gender?

Further Exploration:

A detailed look at MIT’s Tenga learning system: http://robotic.media.mit. edu/portfolio/social-assistive-robots/ A detailed look at Milo: http://www.robokindrobots.com/robots4autism-home/

Fin