View source for Something to do with social robots

{{StudentProjectTemplate
|Summary=Robot Exercise Coach
|Programme=Stub
|Keywords=Social robot, human-robot interaction, recognition, behavior
|TimeFrame=Jan-July 2019
|References=Fasola, J., & Matarić, M. J. (2013). A socially assistive robot exercise coach for the elderly. Journal of Human-Robot Interaction, 2(2), 3-32.
http://delivery.acm.org/10.1145/3110000/3109710/p3-fasola.pdf?ip=194.47.19.128&id=3109710&acc=OA&key=74F7687761D7AE37%2ED64FD9DC22ECC16B%2E4D4702B0C3E38B35%2E6D218144511F3437&__acm__=1540547143_b8ef87a85e37d78837ed72cdc958a5e1

Gold, K., & Scassellati, B. (2006). Learning acceptable windows of contingency. Connection Science, 18(2), 217-228.
http://www.cs.yale.edu/homes/scaz/papers/Gold-ConnSci-06.pdf
|Prerequisites=Should be able to code in python, work with Robot Operating System, and be interested in robots, computer vision, human-robot interaction.
|Supervisor=Martin Cooney, Eren Erdal Aksoy, Fernando Alonso-Fernandez, 
|Level=Master
|Status=Open
}}
Details will be provided later (based on discussion between supervisors and students, to find what is most interesting for everyone involved), but this project will have something to do with social robotics.
And it would be nice to work with a company like Sony Mobile and/or Meta Bytes (I am trying to get their contact info right now).
Someone else could also be involved like Fernando or Eren.


For example: 
*OpenPose for robot exercising.
Baxter (on Ridgeback) will lead a person or group of people to exercise.
Poses will be recognized with OpenPose, and we can maybe do something with deep learning.
An interesting point could be in adapting to interruptions (maybe the people cannot follow along sometimes, so the robot should stop/slow down sometimes and regain control) and/or recognizing exercises for dementia (they can be different than for young adults, like turning a screw).
This class could be at the university gym if we get permission (I have received some contacts and will check). 


*Or, OpenPose for robot first aid. 
Baxter will carry out the Glasgow Coma Scale assessment to determine how conscious a patient is.
This will involve using computer vision to detect where a robot should touch a person to elicit a reaction, then classifying the way a person moves in reaction (e.g., decerebrate, decorticate).
I think this kind of ability to save lives and give real value will be very important for bringing robots into real homes.)

Focus: software
Outcomes: a thesis, demonstrator, video, a local conference paper.