A developmental approach to trust in autonomous systems

The goal of this project is to enable humans and robots to learn to trust each other – by adapting to the socio-communicative level of their partner. And, so, effectively collaborate with each other.
Description of the Project: 

For humans to collaborate effectively and be trusted on shared tasks in human-machine teams, we need to be able to develop [over time] a working relationship. To do this, partners need to have i) a mutual understanding of the task and ii) understand what the other party does and does not know iii) appreciate what the other party can and cannot do. This is known as the ‘Double Empathy Problem’ in the autism/developmental psychology research (Milton, 2012). 

As robotic systems become more human-like and autonomous, this relationship with humans becomes more important; because once established, human-machine teams will need to be more efficient and more robust to be able to safely handle problematic situations, for example, when failure occurs, be it on the system or human side.  

This project will take into account the human’s social functioning level with respect to human-machine teaming by taking individuals’ baseline psychometric measures. One way of measuring these levels in operators is though the validated Autism-Spectrum Quotient (AQ) questionnaire. This project will explore the hypothesis that those higher on the AQ scale will have a mental model of the robot that will be less influenced by the manipulation of human-robot interaction such as dialogue features and touch, as shown with human-human interaction. This will allow for safer human-like robotic behaviour that will adapt to the user’s social and communication skills level. 

Additionally, this project will look at how trust is established, maintained, broken and then repaired as a the relationship between the human-machine teams develops over time over time. 

Resources required: 
Robot for expressive interaction, e.g. the FurHat or Emys
Project number: 
First Supervisor: 
Heriot-Watt University
Second Supervisor(s): 
First supervisor university: 
Heriot-Watt University
Essential skills and knowledge: 
Machine learning
Desirable skills and knowledge: 
ROS, interest in dialogue and NLP Background in Psychology

Buckingham, G., Michelakakis, E.V., & Rajendran, G. (2016). The influence of prior knowledge on perception and action: relationships to autistic traits. Journal of Autism and Developmental Disorders, 46, 1716–1724. 

Crompton, C. J., Ropar, D., Evans-Williams, C. V., Flynn, E. G., & Fletcher-Watson, S. (2019). Autistic peer-to-peer information transfer is highly effective. Autism. 24(7), 1704–1712 

McKenna, P. E., Glass, A., Rajendran, G., & Corley, M. (2015). Strange words: Autistic traits and the processing of non-literal language. Journal of autism and developmental disorders, 45(11), 3606-3612. 

Milton, D. E. (2012). On the ontological status of autism: the ‘double empathy problem’. Disability & Society, 27(6), 883-887.l Disorders, 45, 3606- 3612.