https://upload.wikimedia.org/wikipedia/commons/e/e6/Flickr_-_Official_U.S._Navy_Imagery_-_Doctors_perform_surgery_together..jpg
MASTER THESIS

MHF1 - HUMAN FACTORS IN ARTIFICIAL INTELLIGENCE SUPPORTED CONFORMITY ASSESSMENTS (DEKRA + INTERGO)

SUPERVISOR: PROF.DR. WILLEM VERWEY

35EC

Problem statement

X-ray images are made to assess the conformity of welds in steel with applicable standards and regulation. The interpretation of these X-ray images can be supported with an AI model that detects and classifies welding defects. This could improve the assessment process to be more reliable (fewer defects missed, fewer false calls), more efficient (lower overall cost) and/or have a shorter throughput time (earlier detection can reduce the impact of a defect). Developing an AI model for such an application is however only one step in the path to success.

Whereas AI can cover anticipated defects in an efficient and reliable way, a human assessor can classify unexpected defects in a flexible way. For many situations the combination of AI and a human assessor might be the best solution. Such a combined set-up leads to questions how to account for human factors

Your assignment

                                                                                                                


Question           

  • Which way of processing assures best results in conformity assessment of welds in steel, parallel or serial processing? Provide evidence both from the conformity assessment perspective as well as the human factors perspective.
  • How to keep a human assessor in a productive and reliable state when supported by AI?
  • What are implications for further development?

Method

  • Definition of a theoretical model about interaction between the human assessor and AI
  • Experiment with serial and parallel processing by the AI and human assessor

Sufficient language proficiency in both Dutch and English is required. You will be collaborating with representatives of DEKRA in the Netherlands, Finland, Germany and perhaps other countries.

Results

MSc thesis; publication and/or scientific paper; at least executive summary in English; presentation to DEKRA executives; posts on social media.

Our offer

An inspiring and challenging assignment for a master thesis. The problem statement is founded in the practice of 2 leading companies. Fulfilment of this assignment will need a sound scientific approach; results will have impact on real life practice of both companies.

Date and location

Start: October 2020
Duration: approximately nine months
Location: Utrecht and/or Amersfoort, some interaction with experts abroad

Supervision

Your UT supervisor will be prof. Willem Verwey.
Locally, you will be supervised by representatives of both companies:

DEKRA
Maarten Robers, MSc, MBA  maarten.robers@dekra.com
+31 6 53 43 12 24

INTERGO
M.P. Zeilstra, MSc, Eur.Erg.  zeilstra@intergo.nl
+31 6 54 92 33 07

About DEKRA

DEKRA is one of the world’s leading independent expert organizations. 44 000 qualified DEKRA experts work for safety on the road, at work and at home. The company has its activities on several different areas, among which Product Testing & Certification, Business Assurance, Material Testing &   Inspection, Industrial Safety, Process Safety, Rail and Claims & Expertise.
www.dekra.nl or www.dekra.com

About INTERGO

INTERGO is a leading consultancy in safety, ergonomics and human factors. INTERGO has more than 50 years of experience internationally. We apply our deep knowledge of human behaviour to enable companies to perform safe and successfully. We take our social responsibility by actively contributing to scientific research in the field of safety, ergonomics and human factors.                        
www.intergo.nl or www.intergo.eu

MHF2 - FRONTAL EEG POWER IN THE THETA BAND AS A MEASURE OF MENTAL WORK LOAD

SUPERVISORS: DR. ROB VAN DER LUBBE,  DR. SIMONE BORSCI

35EC

In two recent EEG studies, we examined whether frontal theta power is a proper index of increased mental work load. In a Sternberg task, we varied working memory load, and participants had to keep either 1, 2, or 4 digits in mind. Upon presentation of a probe digit, participants had to indicate whether the probe was part of the memory set or not. In another task, the ADD-N task, participants had to add 0, 1, or 2, to each number of a four-digit number. The latter task is also thought to reflect increased mental effort (Kahneman, 2011). To validate frontal theta as an index of mental work load, results in both tasks should show comparable effects across the same individuals. Goal of the study is to examine whether these tasks are indeed measuring comparable effects.

MHF3 - BENEFITS OF EXTENDED MULTISENSORY SPACE DURING AUTOMATED DRIVING: A DRIVING SIMULATOR STUDY

SUPERVISORS: DR. FRANCESCO WALKER, PROF. DR. ING. VERWEY

35EC

Achieving safe transfers of control is a key goal for both Conditional and High automated driving. Right now, as soon as the driver gets out of the loop and therefore stops paying attention to the road, safe transfers of control cannot be guaranteed. Our study will point to new ways of keeping the driver “connected” to the vehicle, allowing smoother and faster transfers of control.

The proposed study will investigate the influence of visuorespiratory synchronization on drivers’ feelings of safety and take-over reaction times. We hypothesize that visuorespiratory synchronization will induce an extension of participants’ peripersonal space (i.e., the space immediately surrounding the body), influencing their trust and reaction times when asked to manually take back control of a simulated automated vehicle.

If you want to know more about this project, please send an email to Francesco (f.walker@utwente.nl).

MHF4 - ADAPTING AUTOMATED VEHICLE BEHAVIOUR TO USER TRUST: A DRIVING SIMULATOR STUDY

SUPERVISORS: DR. FRANCESCO WALKER, PROF. DR. ING. VERWEY

35EC

Trust is one of the main factors slowing down the adoption of automated driving technology. This study aims to improve trust by tailoring automated vehicle (AV) behaviour to each user. In the driving simulator of the UT, while being driven by an AV, participants will continuously report how much they trust its behaviour. They will indicate their trust through a slider, with “0” indicating “No trust” and 100 indicating “Full trust”. An alert will be sent to the experimenter if, for 30 seconds straight, the participant’s trust will be 10% below or above the previously reported value. The experimenter will decrease or increase AV’s speed by 5 km/h every time an alert is received. We hypothesize that, when compared to two control groups (Control1: speed stays constant; Control2: speed randomly changes), adapting vehicle’s speed to the user’s trust will lead to higher trust levels in individuals that tend to distrust automation.

If you want to know more about this project, please send an email to Francesco (f.walker@utwente.nl).

MHF5 - ARTIFICIAL INTELLIGENCE CONVERSATIONAL AGENTS: A MEASURE OF SATISFACTION IN USE

SUPERVISOR: DR. SIMONE BORSCI

35 EC

Background

Conversational agents, such as chatbots and voice interfaces, can be used for multiple purposes e.g., support customer experience with services etc. These new tools are growing and more and more integrated into systems such as websites, social networks, cars. Smart and AI-based conversational agents are shaping the future of human-computer interaction however little is known about how to assess people reaction and satisfaction after the use of these systems.

Goals

Advance previous work done on a new scale to assess satisfaction with chatbots. Your experimental work will focus on the evaluation of conversational agents to further streamline the reliability and validity of the scale.

Your work will consist of testing with a remote usability test different chatbots with a set of tools, including the new scale to perform a confirmatory factorial analysis. You should be aware of statistical methods regarding factorial analysis and be able to use R. The target is to involve at least 100 participants working (potentially) in a team.

This assignment is suited for a team of two master students.

Key references

  • Coperich, K., Cudney, E., & Nembhard, H. Continuous Improvement Study of Chatbot Technologies using a Human Factors Methodology.
  • Duijst, D. (2017). Can we Improve the User Experience of Chatbots with Personalisation? MSc Information Studie, Amsterdam. 
  • Følstad, A., & Brandtzæg, P. B. (2017). Chatbots and the new world of HCI. interactions, 24(4), 38-42.
  • Hill, J., Ford, W. R., & Farreras, I. G. (2015). Real conversations with artificial intelligence: A comparison between human–human online conversations and human–chatbot conversations. Computers in Human Behavior, 49, 245-250.

Thesis + Internship


MHF6 COLLABORATION AT A DISTANCE: REMOTE EXPERT SUPPORT DURING A TROUBLESHOOTING TASK (TNO)

SUPERVISOR: PROF.DR. JAN MAARTEN SCHRAAGEN

25 + 10 EC

Due to increasing pressures for more efficiency, there is increasing interest in remote, collaborative work, for instance in troubleshooting. An expert troubleshooter would not have to be physically present on location, and could provide support to someone less expert (a ‘novice’) via remote communication channels, using audio, video, or Augmented or Virtual Reality. Cost savings could occur in case such an expert could assist multiple novices at different locations simultaneously, without having to be physically present him-/herself.

There are several different work fields that could benefit from an expert that provides remote support to novices on location. We are particularly interested in remote work for platforms of the Royal Netherlands Navy. By keeping the expert remote in a type of control centre, they will be able to help out on several different issues occurring in different platforms at the same time. A possible problem that could arise from this approach is that it would create a new generation of employees that are not truly experts at troubleshooting. One possible research question may be: What is the effect on a novice’s learning curve when he or she performs the troubleshooting task either supported by a remote expert or on their own with a troubleshooting manual? Other research questions are possible as well.

This research is carried out in conjunction with TNO, although the research will in all likelihood, given the current COVID-19 measures, be carried out at a distance.

This is a 25 EC master thesis assignment that may be complemented with a 10 EC internship.

TNO offers an exciting state of the art research environment, where you will be part of a larger team working on control room design. TNO offers a monthly monetary compensation.

Due to the nature of the research position, only students with the Dutch nationality can apply.

For more information about the department at TNO where you will be working, see:

https://www.tno.nl/en/focus-areas/defence-safety-security/expertise-groups/human-behaviour-and-organisational-innovation/

For more information, please contact:

UT Supervisor: Prof. dr. J.M.C. (Jan Maarten) Schraagen, j.m.c.schraagen@utwente.nl

Supervisor at TNO: Tom Hueting, tom.hueting@tno.nl