The Wizard behind the Robot

Case Reports

Wizard of Oz studies with Children

Hoysniemi J & Read J

Link to article: citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.124.4282&rep=rep1&type=pdf

What? This article reports on case studies about Wizard of Oz robots and children. ‘Wizard of Oz’ refers to the human behind the curtain who animates robot’s interaction with children. The robot is not autonomous but operated to some greater or lesser degree by a human wizard, as close to real time as possible.

Why? Wizard of Oz studies are interested in modelling and examining characteristics of computer-human interactions and the capabilities and limitations of autonomy in robots. These studies allow researchers to efficiently try out ways of interacting that might become programmed behavior in a more autonomous robot. It allows collection of data.

What was found? What is the balance between robot autonomy and the wizard in responses to the participant, and the delay inherent in this. Delays that occur as the wizard prepares a response for the robot to make were recognized as a limitation. Of mention is the notion of deception: whether or not the wizard being hidden or seen was an issue for children. Children mis-attributed autonomy to the present wizard even when there were keyboarding sounds occurring at the same time as the wizard was responding. Is there a responsibility on researchers regarding consent when children don’t understand there is a human behind the robot?

My (robot) friend

Blog post

Making new robot friends: Understanding children’s relationships with social robots

Westlund JMK

Published in mit media lab: Personal Robots Group blog, June 13, 2017

Link to blog: media.mit.edu/posts/making-new-robot-friends/

"Hi, my name is Mox! This story begins in 2013, in a preschool in Boston, where I hide, with laptop, headphones, and microphone, in a little kitchenette. Ethernet cables trail across the hall to the classroom, where 17 children eagerly await their turn to talk to a small fluffy robot."

In this engaging blogpost, the author describes a robot used to explore the nature of relationships children build this social, teleoperated machine. In playtests with children, the author learns they form an understanding of robots that differ from the relationship with the robot’s human operator; from pets, from inanimate possessions and from people. Adult ontologies are not applied by children: person or machine, real or imaginary; no condition excludes the possibility of a friend. Characteristics like expressiveness and responsiveness are found to contribute to connection and learning. Robots can ask questions that encourage the child to greater engagement in a story and cue parents to ask similar questions.

Finally, the author expresses concerns about the ethics of providing engaged machines to children and issues a call to engagement in ‘roboethics’.

Research proposal: Transparency, Teleoperation, and Children’s Understanding of Social Robots, Westlund JMK & Breazeal C

Link to research proposal ieeexplore.ieee.org/document/7451888/

Robot rules and features

Survey

Designing a Robot Companion for Children with Neuro-Developmental Disorders

Gelsomini M, Degiorgi M, Garzotto F, Leonardi G, Penati S, Ramuzat N, Silvestri J & Casadonte F

IDC '17 Proceedings of the 2017 Conference on Interaction Design and Children 451-457

DOI>10.1145/3078072.3084308

Link through to full text: //dl.acm.org/citation.cfm?id=3084308

Objective: The authors set out to do two things.

1. to collect information from 15 pediatric therapy ‘specialists’ and create a list of desired properties in a robot companion
2. to study the classes of features that are present in available robots

Process: 21 common robots are chosen, and the concepts are both outlaid in matrixes for comparison.

Findings: The authors come up with “Robot’s Rules”. First, desired properties: a therapy robot should facilitate, prompt, restrict, emulate, provide feedback; and act as a social mediator, an affective agent and an emotional agent. Secondly, the five classes of features that social robots should have are identified: appearance, multimodality, multisensorialilty, configurability and safety.

Sensorimotor Robot Therapy: the Anklebot

Pilot study

Pediatric robotic rehabilitation: Current knowledge and future trends in treating children with sensorimotor impairments

Michmizos KP & Krebs HI

Neurorehabilitation (pre-press) DOI: 10.3233/NRE-171458

Link to abstract: www.ncbi.nlm.nih.gov/pubmed/28505989

Objective: "As long as motor learning remains the major working hypothesis for sensorimotor rehabilitation, a well-designed robotic therapy… should follow the principles of motor learning, namely massed practice, cognitive engagement and functional relevance." Sensorimotor therapy is identified as three elements: discrete and rhythmic movements, and interaction with the environment.

Process: The application of the Pedi-Anklebot 'robot' is to target ankle movements and functional walking for children with motor disabilities. The authors use the games designed to treat reduced speed, strength, accuracy, co-ordination; and cognitive or perceptual difficulties. To provide comprehensive sensorimotor therapy, practice included discrete sub movements, rhythmic oscillations and practice with ground contact.

Conclusion: "To fully harness the therapeutic power of adaptation, we need to continue our research on the special characteristics of [the] sensorimotor control… on the differences between rhythmic and discrete movements, as well as mechanical impedance training."

Robots Learn to Play

Review

Robots Learn to Play: Robots Emerging Role in Pediatric Therapy

Howard AM

Proceedings of the 26th International Florida Artificial Intelligence Research Society Conference, 2013

Link to full text: www.aaai.org/ocs/index.php/FLAIRS/FLAIRS13/paper/view/5838/6033

Objective: Therapy for children includes play; at best, robots will express this quality as they are brought into the pediatric setting. Three types of robots are discussed in this article: robots (smart toys) that are operated by alternative inputs such as the smart phone, robotic orthoses that include an element of play, and autonomous robots that interact through play. In each case, examples are given.

Discussion:

1. ‘Robot’ or smart toys are considered here as toys which children can play with using an alternative means of manipulation. True smart toys possess some awareness of their surroundings and additionally require some degree of autonomy to meet the definition of robot.
2. Robotic orthoses or exoskeletons guide, limit and assist in motor movements of upper and lower extremities. Play can be incorporated into this therapy.
3. Autonomous robot toys which engage children with intentional, purposeful imitation play. "While typically developing children possess the ability to imitate others from birth, children with pervasive developmental disorders, such as autism, demonstrate significant difficulty in object and motor imitation. Imitation skills are thought to be closely related to early language and social abilities."

Conclusion:"It seems natural then that this research thread (autonomous robotic playmates capable of engaging children in shared manipulation-based play), along with … quantitative results… emerges as the next step in the domain of robots for pediatric therapy."

Balanced Games = More Fun

Pairs study

How Game Balancing Affects Play: Player Adaptation in an Exergame for Children with Cerebral Palsy

Hwang S, Schneider ALJ, Clarke D, MacIntosh A, Switzer L, Fehlings D & Graham TCN

Conference proceedings: ACM DIS 2017, June 10 – 14, 2017, Edinburgh UK

Link to abstract: dx.doi.org/10.1145/3064663.3064664"

Player balancing changes game mechanics to accommodate differences in players’ manual and cognitive abilities. Addressing the imbalances also involves players’ awareness of a new fairness that may motivate both the advantaged and disadvantaged players.

Objective : When children with CP play knowingly balanced exergames, do the effects (including motivation) persist?

Process : Eight participants 8 – 14 years old with GMFCS scores of II or III experienced 6 hour-long sessions of either balanced or unbalanced game play.

Findings : "Our results showed that motivation was higher in balanced versus non-balanced conditions [and] perceived fun and fairness were higher for both winners and losers in balanced versus non-balanced conditions."

Testing a Therapy Assistant Robot

Proof of Concept

Adapting a General Purpose Robot for Paediatric Rehabilitation: In-situ Design of a Socially Assistive Robot

Carrillo FM, Butchart J, Knight S, Scheinberg A, Wise L, Sterling L & McCarthy C

Link to article:https://arxiv.org/abs/1705.05142

Objective: Socially Assistive Robots are engaging and motivating. For social robots in pediatric rehabilitation, what are the therapist, carer and patient-centric properties needed for robots to be effective stand-alone therapeutic aids? What are the lessons from ongoing use in a clinical setting?

Process: Development began with weekly visits to clinic with the robot to stimulate brief discussions with parents and therapists; and interactions with children, followed by programmers involving therapists in cycles of programming the NAO robot and reviewing the exercise sequences. Later development shifted the sole operation of the robot to therapists, parents and care-givers. Clinical use of the robot had 5 therapists working with 9 patients over 14 sessions and continued the iterative design process.

Outcomes: The authors state: “this (deployment) approach has led to a system that not only meets minimum operational and therapeutic requirements for clinical deployment, but also has clearly established priorities for further development as we prepare for formal clinical trials of the socially assistive robot for pediatric rehabilitation.”

Distance Rehab for Kids – A Case Report

Case Report

Feasibility of Pediatric Game-Based Neurorehabilitation Using Telehealth Technologies: A Case Report

Reifenberg G, Gabrosek G, Tanner K, Harpster K Proffitt R & Persch A

American Journal of Occupational Therapy, 71 7103190040

DOI: doi.org/10.5014/ajot.2017.024976

Objective: Feasibility of and suitable outcome measures for distance-monitored video game rehabilitation.

Process: A child with hemiparetic spastic cerebral palsy played 7 hours weekly for 8 weeks of computer games monitored at a distance by a therapist via videoconference technology KUBI The setup permitted the remote therapist to move their camera (iPad) to see the child and the child’s position. The distance therapist consulted with child and parents 30 minutes each week for technical issues.

The outcome measures that were expected to detect change before and after the test condition were: Quality of Upper Extremity Skills Test (QUEST), Bruininks-Oseretsky Test of Motor Proficiency 2 Ed.(BOT-2), Assisting Hand Assessment (AHA), Pediatric Evaluation and Disability Inventory-Computer Adapted Test (PEDI-CAT) and Pediatric Motor Activity Log (PMAL). Pre- and post- tests were applied two weeks before and after the game sessions.

Findings: Feasibility – the audio quality was sufficient for the therapist to interact with the child and family, and the video quality allowed the therapist to see the child’s activity without seeing the TV monitor. There were no insurmountable technical issues though this subject and family were familiar with the technologies used and so this case did not represent a truly novel situation.

Measures – All the measures were sensitive to change in the child’s performance before and after the test condition except the QUEST.

“The use of telehealth technologies provides practitioners with a mechanism to supervise treatments for clients in underserved communities. This research provides initial evidence that it is feasible to administer game-based neurorehabilitation and telehealth technologies and monitor relevant outcomes.”

Social Robots: NAOTherapist

Proof of Concept

NAOTherapist: Autonomous Assistance of Physical Rehabilitation Therapies with a Social Humanoid Robot

Pulido J C, Gonzalez J C & Fernandez F

Proceedings of IWART 2016: International Workshop on Assistive and Rehabilitation Technologies, Elche, Spain, December 14 – 16, 2016

Link to paper

Link to video

Objective: The goal of this work is to identify the architecture needed for NAO to provide therapy without human supervision.

Discussion: Social robots may act to relieve the workload of the therapist, fulfilling the clinical objectives autonomously over many treatment sessions. In this study, early requirements of a robotic therapist have been previously identified: the children had no problem following the instructions; they enjoyed the activity and adhered to the regimen.

For autonomous therapy, the robot must possess social awareness in addition to offering exercises that it models, monitors, encourages and corrects. It should be able to report back results to the therapist. In this case, NAO is doing successive arm poses with the child. The robot greets the child. It then models each arm pose, checks on the position achieved by the child, encourages and corrects the pose by using a variety of strategies to model the pose and to show how move into that pose. NAO times the correct achievement of the pose. Here, Kinect is being used separately to capture the poses for feedback to the therapist. NAO indicates the completion of each pose and the exercise routine and offers a wish to "play again tomorrow".

Robot assists with kids mobility

News

Interactive robot to promote rehabilitation for children with special needs

Kukich D. Published in UDaily, Communications and Public Affairs, University of Delaware, Jan 11, 2016

Link to story

At the University of Delaware in Newark, a research team including mobility researcher Cole Galloway are using "NAO", an interactive human-like robot in a new approach to pediatric rehabilitation. Dr. Galloway, who is known for his Pediatric Mobility Studio and GoBabyGo program, is incorporating the robot into his lab research. His research is directed by the concern that children with motor disabilities have needs that are not met by the training and equipment available today. "Young children’s overall knowledge depends on their ability to be mobile with peers — once they start moving, they begin to learn about the world in fundamentally different ways." The intention of including NAO in the mobility environment is to have the children engage with the robot, and for the robot to facilitate training.

Links

Update on Dr. Galloway's activities

Pediatric Mobility Lab and Design Studio

GoBabyGo "All people exploring their world via independent mobility!"

NAO robot - Softbank Robotics.com

Gait Robots Help Upper Body Posture

Randomized Controlled Trial

Robotic-assisted gait training improves walking abilities in diplegic children with cerebral palsy

Wallard l, Dietrich G, Kerlirzin Y & Bredin J

European Journal of Paediatric Neurology 2017, http://dx.doi.org/10.1016/j.ejpn.2017.01.012

Link to abstract

Objective: For children with diplegic cerebral palsy, robotic-assisted gait training (RAGT) can result in improved gait and posture compared to children who receive only traditional physical therapy. Of particular interest here is dynamic equilibrium control: upper body strategies such as shoulder elevation and elbow flexion used to maintain balance while propelling forward.

The authors propose: "RAGT presents beneficial effects on improvement of postural and locomotor functions of the patient resulting in a reorganization of gait pattern and full-body kinematic illustrating the dynamic equilibrium control in gait. This would translate in a better stabilization of the head, a better control of the displacement of the arms associated with an improvement in the kinematics of the lower limb."

Process: Thirty children with diplegic cerebral palsy from 8 to 18 years of age were randomly assigned to either a treatment group or a control group. The treatment group received 40-minute sessions of Lokomat Pediatric robot-assisted walking therapy, 5 days a week for 4 weeks. Body-weight support and walking speed were altered through the sessions to match the child’s functional capacity. The control group received daily 40-minute sessions of traditional therapy.

Findings: Gross Motor Function Classification System dimensions GMFM- D (standing) and GMFM-E (walking/running/climbing) were measured and showed improvement within the treatment group, and between treatment and control groups.

Children in the treatment group demonstrated new strategies for gait with significantly reduced shoulder elevation and elbow flexion, "more appropriate control of the upper body... associated with an improvement of the lower limbs kinematics which is similar to the values observed in typically developing children."

There was no follow-up to indicate whether the effects are preserved over the long term.

Robot Gait Rehab: Systematic Review

The effectiveness of robotic-assisted gait training for paediatric gait disorders: systematic review

Lefmann S, Russo R & Hillier S

Journal of NeuroEngineering and Rehabilitation 201714:1

DOI: 10.1186/s12984-016-0214-x

Link to full text: jneuroengrehab.biomedcentral.com/articles/10.1186/s12984-016-0214-x

Objective: to review evidence for robotic-assisted gait training (RAGT) in children and adolescents with gait disorders as a consequence of cerebral palsy; or of musculo-skeletal, neurological orthopedic origin.

Process: Six databases were searched from 1980 to October, 2016. Data were extracted using the PRISMA model. http://www.prisma-statement.org/

Outcomes: Three trials that met the standard of randomized controlled trial; two were selected. Meta-analysis of the two trials showed "weak and inconsistent evidence regarding the use of RAGT for children with gait disorders".

Studies of lower design related mostly to children with cerebral palsy; these did show improved gait attributes such as speed, standing ability and walking distance. "… a clinical recommendation to adopt RAGT in children and adolescents with gait impairment cannot be made until more consistent findings are reported in larger randomized controlled trials."

Robot-Assisted Rehabilitation

Review

The present and future of robotic technology in rehabilitation

Laut J, Porfiri M & Raghavan P.

Curr Phys Med Rehabil Rep Nov 2016

DOI: 10.1007/s40101-016-0139-0

Link to abstract: link.springer.com/article/10.1007/s40141-016-0139-0>

Robot-assisted rehabilitation can contribute to conventional therapy by delivering consistent training and by benchmarking and monitoring progress. Robots can offer efficiencies in time and over distance. They may act passively or actively, and may be “end-effector robots” which act through an handle that is gripped or provide an exo-skeleton that mimics the movements of the limb. Robot-assisted rehabilitation can function remotely through telerehabilitation, reaching clients in their homes.

Robot therapies leverage entertainment software to provide motivation and engagement with an aim to boost compliance with interventions, and, through distance technologies, may offer efficiencies to the therapist. Equipment costs and the need for therapist-friendly software that targets rehabilitation intervention goals remain challenges to wider adoption.

Amadeo robot for kids rehab

Hand Robotic Therapy in Children with Hemiparesis

Pilot Study

Bishop L, Gordon AM & Kim H

American Journal of Physical Medicine and Rehabilitation 2016;00,00-00

DOI: 10.1097/PHM.0000000000000537

Objective: In a sample of children with hemiparetic upper extremity cerebral palsy, is the Amadeo Hand Robot System feasible for retraining grasp and release?
Amadeo Hand Robot System

Link to abstract: www.ncbi.nlm.nih.gov/pubmed/27386804

Method: 12 participants, 6 – 17 years old with upper extremity hemiparesis were measured initially, after a 6 week pre-intervention stage and following 6 weeks of three- hour long sessions. The Assisting Hand Assessment was chosen as the primary measure.

Findings: Participants tolerated the Amadeo, and the Assisting Hand Assessment showed significant change pre- to post-test. The Jebson-Taylor Test of Hand Function, the Upper Extremity Fugl-Meyer and the Quality of Upper Skills Test all revealed non-significant positive change; and the Pediatric Evaluation of Disability Inventory showed no change.

What qualities should a rehab robot have?

Pilot study

Robots in Rehab: Towards socially assistive robots for paediatric rehabilitation

McMarthy C, Butchart J, George M, Kerr D, Kingsley H, Scheinberg A M & Sterling L

OzCHI '15 Proceedings of the Annual Meeting of the Australian Special Interest Group for Computer Human Interaction, pp. 39-43

Link to abstract: dx.doi.org/10.1145/2838739.2838791

Objective: What are the possible clinical benefits of social robots in rehabilitation for children? This preliminary study explores the characteristics required of robots which interact with children in a post-surgery inpatient setting.

Process: In this setting, NAO robot was used to mimic exercises and motivate children when the therapist was not present. Children had two sessions weekly with a therapist, followed by a third session with NAO, over a period of five months. NAO speaks and demonstrates each exercise, provides a focus during the exercises to minimize boredom, provides encouraging and rewarding vocalizations and records the child’s activity for later review.

Findings: Outcomes of interest are specific to the characteristics of the device. Issues that detracted from NAO’s functionality were movement and speech slowness, physical stability and limited balance, voice clarity and battery life.

Therapy Robots: What Therapists Think

Survey

Views of Physiatrists and Physical Therapists on the use of Gait-training Robots for Stroke Patients

Journal of Physical Therapy Science 28: pp. 202 – 206, 2016

Kang, CG, Chun, MH, Jang, MC, Kim, W & Do, KH

Link to full text: www.jstage.jst.go.jp/article/jpts/28/1/28_jpra-2015-652/_article

Objective: to survey therapists on the desired characteristics of gait-training robots.

Process: 100 physiatrists and 100 physical therapists from 38 hospitals were given questionnaires. One-third of the respondents had worked with a gait-training robot before, and a third had more than 5 years experience with stroke rehabilitation.

Results: therapists thought that treadmill type gait-training robots should be used with non-independent ambulators to "improve treatment effects" one to three months after stroke. Sensors should detect electromyography, foot pressure, joint torque, tilt sensor. Qualities of design and production should include: stability, comfort, easiness on-and-off, "fitting sense" and light weight.

Effectiveness of robot play for kids with CP

Case Series

Playfulness in Children with Limited Motor Abilities when using a Robot

Rios-Rincon, A. M., Adams, K., Magill-Evans, J. & Cook, A.

Physical & Occupational Therapy in Pediatrics, 00(00):1 – 15, 2015

Link to abstract: www.ncbi.nlm.nih.gov/pubmed/26566226

Objective: For children with cerebral palsy, will robot play increase playfulness? Children with cerebral palsy may have fewer opportunities to play. “Promoting playfulness in children with CP… might have a positive effect on self-determination, adaptive behaviors, and problem solving in creative ways. "Will Test of Playfulness (ToP) scores increase during the intervention?"

Process: Four 4 – 9 year old children with Gross Motor Function Classification System (GMFCS) levels 4 or 5 and Manual Ability Classification Scale (MACS) level 4 or 5 and they needed to be able to respond to yes/no questions and to operate the robot using two switches. During the intervention phase, the children were encouraged by their mothers to play with the robot for 15 minutes twice a week for 5 weeks.

Findings: The baseline ToP scores for three of the four children were stable. During the intervention, all four children showed significantly increased scores. "This study adds to the limited evidence for the effectiveness of interventions for play… (and) that adapted Lego robots promote playfulness in children with cerebral palsy who have limitations in mobility and manual ability through free play in family routines."

Robot therapy for children’s arm rehabilitation

Effects of robotic therapy on upper-extremity function in children with cerebral palsy: A systematic review

Chen, Y-P & Howard, A. M.

Development Neurorehabilitation, Early Online 1-8, 2014
DOI: 10.3109/17518423.2014.899648

Link to abstract: www.ncbi.nlm.nih.gov/pubmed/24724587

Objective: To systematically review the evidence of robotic therapy for the effects on upper extremity function for children with cerebral palsy.

Process: Robotic therapy is understood here to be “the application of electronic, computerized control systems to mechanical devices designed to perform human functions”. A literature search resulted in nine articles (7 case studies, one single case study and one randomized crossover design) relating to three robotic systems; CosmoBot, InMotion2, and NJIT-RAVR. These were reviewed on 11 qualitative scales, with results ranging from 3/11 to 9/11 and averaging 6/11, using a protocol developed by Kwakkel et al and Cambach et al. The studies’ outcome measures were grouped using the ICF model: range of motion, muscle tone and kinematics outcomes were classified as either body structure and function; or activity outcomes, which were scored on the QUEST, Melbourne or Fugl-Meyer scales; or on parent questionnaires. Effect sizes were calculated using the Cochrane Review Manager.

Results: Studies all reported either increased reach duration, smoothness, reduced tone or improvement in some clinical measure. The majority of case studies in this systematic review limit the recommendations that could come from more robust, group design studies. No recommendations can be made.