SUPERVISORS: A. BOELHOUWER, W.B. VERWEY
The majority of driving studies are performed within controlled environments such as driving simulators. A driving simulator has multiple advantages that make it attractive to use compared to driving studies on the real road. The main advantage is that dangerous situations can be studied without the safety risks. Furthermore, situations can be manipulated and repeatedly tested in a controlled way that would be very difficult to achieve on the road. The validity of this method has been studied for a long time (Kaptein, Theeuwes, & Horst, 1996; Wang, et al., 2010). However, due to practical considerations such as time and costs, driving studies use other methods instead. For example, videos of driving situations may be distributed through online surveys. This is mainly done when a response or judgment is necessary from a large amount of participants within a limited amount of time. It is however not clear how the results of driving studies may be affected by the use of these types of methods.
In a previous study, we have used video recorded driving sessions within a physical driving simulator. It is necessary to see how the same experiment translates through other platforms. Are the results similar through different implementation forms? The aim of this study is to compare and analyze different methods of driving studies. One example of these methods is the presentation of video recorded driving sessions in a physical driving simulator. Another method could be the static presentation of an image in an online survey. What are the different methods that may be used for driving studies, and what are the individual benefits and drawbacks of these methods?
You will extend our previous driving simulator study on information presentation for automated driving to other platforms besides the driving simulator (for example through online surveys). This includes the preparation of the experiment within these platforms and setting out the actual experiments. Afterwards you will analyze and compare the results from the different platforms.
The project can be adjusted to fit the goals and interests of the master student. The project will be carried out at the UT. For more information on this project, please contact email@example.com
SUPERVISORS: A. BOELHOUWER, W.B. VERWEY
Starting 2015, large car companies have started testing and rolling out car systems that combine multiple automated systems, to control both longitudinal and lateral behaviour simultaneously. Examples include the Autopilot by Tesla (Tesla, 2017) and Intelligent Drive by Mercedes-Benz (Mercedes-Benz, 2017). In these cases, the driver still needs to monitor the situation and be able to take back control at any time, which classifies as SAE level 2. The levels of automation by SAE range from 0 (manual driving) to 5 (full automation under all circumstances).
In research and industry, higher levels of vehicle automation and accompanying concepts are developed. Up until level 4, the driver is always the fallback for the system. In level 4 automation, the driver still needs to take back control after a timely take-over request, however in case the driver does not respond the system has a built in fallback procedure. This procedure should bring the vehicle to a standstill in a safe and controlled manner. A simple example for this is turning on the warning lights and slowly deceasing speed.
In the current I-CAVE program, that focusses on developing a dual mode connected vehicle, it is unclear what fallback procedure should be implemented in specific scenario’s. The aim of this study is therefore to investigate and advise on an appropriate fallback procedure that promotes both traffic safety and acceptance of drivers. You will test different fallback procedures in a driving simulator. You will be working closely with the simulator as you both make the scenarios and test participants in here. Thereafter, you will analyze the different fallback procedures and advise the technical development team of the I-CAVE program. The specific use cases of the I-CAVE project will be used as a starting point for your study.
As you will be working with our driving simulator, it is advised to have some experience in basic Python programming skills. The project can be adjusted to fit the goals and interests of the master student. The project will be carried out at the UT. For more information on this project, please contact firstname.lastname@example.org
SUPERVISORS: DR. ROB VAN DER LUBBE, DR. MARTIN SCHMETTOW
Some websites are more easily encoded in our memories than others. Part of this variability may be due to specific features of the website, but fluctuations in the state of our brains will also affect memorability. Measures derived from the EEG, especially power in the alpha and theta band are likely predictors of what will be remembered later on, and what not. An experiment will be performed with a variation of the so-called oddball paradigm in which a standard website, target websites, and deviant websites will be presented for 1,000 ms. EEG will be measured during this oddball session. In a second test, participants will be presented with several websites, half of them presented before, half of them new, and participants have to indicate whether they saw the websites before. Acquired EEG data may reveal whether brain states predict memorability of the websites in the test session.
SUPERVISOR: DR. MARTIN SCHMETTOW, DR. MARLEEN GROENIER
Minimal invasive surgery (MIS, keyhole surgery) is one of the most important developments in surgery during the past decade. So far, MIS training has been following the apprenticeship model, where trainees watch a hundred or so procedures, before putting their own hands on a patient. However, with the emergence of virtual reality training simulators, this is soon going to change. This new technology is promising, as it provides safe and highly repeatable training opportunities, as well as assessment of a person’s capabilities.
Figure 1. Learning curves for the efficiency of motions in laparascopy.
Research has shown that doctors differ in how fast they acquire the necessary skills, and the proficiency they reach in the long term. Figure 1 shows three example learning curves.
These differences can partly be explained by individual differences between doctors, such as amount of training and cognitive capabilities like visual-spatial processing ability.
In fact, minimally invasive procedures are quite demanding:
- there is no perception of depth
- the procedure often require counter-intuitive movements
- haptic feedback is limited
The research field has long been on the quest for assessing whether such abilities are reliable in predicting whether someone is likely to become a good MI surgeon.
MIS simulators also have great potential during the process of training. In every simulator session an abundance of performance variables are recorded (e.g., motion efficiency, see Fig 1). It is compelling to use such data to track the training progress of individual trainees. Furthermore, using learning curves, it may be possible to predict ahead of time, what performance level a trainee is going to reach (for example, looking at Fig 1 participant 06 will reach a much better level than 03). In the future it may be possible to use learning curves from simulator data to select those with high potential for a training program.
In this master project you will set up a suite of simulator tasks and run a learning experiment on novice participants in the field of endovascular surgery (blood vessels, e.g. to treat an angina pectoris).
Based on the results, you will draw conclusions on the span of individual differences and the feasibility of simulator-based assessment. During your project you will work together in a team of two students. You will work in close collaboration with endovascular surgeons at the Medisch Spectrum Twente.
SUPERVISOR: DR. SIMONE BORSCI
Chatbot, intended as computer programs designed to simulate the conversational interaction with people, are increasingly used in the human interaction with digital and online service.
The use of an artificial intelligence assistant, who may answer to end-users 24/7, is a growing trend in digital technology industries, used to support training and better navigation of websites, costumers’ service and experience, or user information seeking (e.g., health and food). There are different approaches to develop chatbots, but each type of chatbot has limitations which may disrupt the interaction. The core common feature and aim of chatbots is to enable a natural language process (NLP), or a more simple Conversational Interfaces (CI) interaction to help end-users to achieve their goals.
Nevertheless, our initial hypothesis is that people (as adaptive agents) could quickly learn how to minimize the wording during the conversation with chatbots by acquiring a basic set of “command lines” to speed up their tasks achievement.
The main objectives of this project is to: i) review the state-of-art on chatbots development and applications, ii) identify potential chatbot services, define a preliminary test to investigate the initial hypothesis, and to explore emerging hypothesis iii) Define a set of hypothesis and an experimental design; iv) Use observational methodologies, questionnaires and eye tracker tools to analyze people ability to adapt their conversational style for a rapid achievement of their aims during the interaction with chatbots, and to gather data about people reactions and experience with chatbots.
Example research questions (not limited): i) is interaction with chatbots a step toward conversational interface or a step back to command line interfaces; ii) which are the limits and the constraints of chatbots experience; iii) When chatbots enable good and bad experience.
List of references related to the topic:
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.
Kuligowska, K. (2015). Commercial Chatbot: Performance Evaluation, Usability Metrics and Quality Standards of Embodied Conversational Agents. Browser Download This Paper.
Xuetao, M., Bouchet, F., & Sansonnet, J.-P. (2009). Impact of agent’s answers variability on its believability and human-likeness and consequent chatbot improvements. Paper presented at the Proc. of AISB.
Kerly, A., Hall, P., & Bull, S. (2007). Bringing chatbots into education: Towards natural language negotiation of open learner models. Knowledge-Based Systems, 20(2), 177-185.
Hirschberg, J., & Manning, C. D. (2015). Advances in natural language processing. Science, 349(6245), 261-266.
Shawar, B. A., & Atwell, E. (2007). Chatbots: are they really useful?. In LDV Forum (Vol. 22, No. 1, pp. 29-49).
McTear, M., Callejas, Z., & Griol, D. (2016). The conversational interface. New York: Springer, 10, 978-3.
Agostaro, F., Augello, A., Pilato, G., Vassallo, G., & Gaglio, S. (2005, September). A conversational agent based on a conceptual interpretation of a data driven semantic space. In AI* IA (Vol. 3673, pp. 381-392).
Heller, B., Proctor, M., Mah, D., Jewell, L., & Cheung, B. (2005, June). Freudbot: An investigation of chatbot technology in distance education. In EdMedia: World Conference on Educational Media and Technology (pp. 3913-3918). Association for the Advancement of Computing in Education (AACE).
Chatbot integrated in websites
Ibrid chatbot with forced answers
Facebook messanger based
Artificial intelligence for conversation
Master thesis with internship
INTERNAL SUPERVISOR: DR. SIMONE BORSCI
WORKFLOW REDESIGN AND TRUST TOWARD ENHANCED PSYCHO-RADIOLOGY PATHWAYS AND TOOLS
1. Summary of the project
New approaches, including advanced neuroimaging techniques supported by artificial intelligence (NeuroAI), are today available to support clinical decision making to select the appropriate treatment for people with neurological and mental disorders. Philips is exploring the opportunity to develop new clinical pathways to efficiently include NeuroAI in to healthcare services.
This project will focus on mapping the current clinical pathways (patient journey, data flow and workflow) with appropriate human factors and ergonomics methods, to inform the decision making of clinicians, and to design future and enhanced pathway with the incorporation of NeuroAI in the field.
2. Student profile and selection criteria:
One internship position is available for students who may apply no later than 09 March 2018 by sending an email application letter to Simone Borsci (email@example.com)
explaining why they want to apply, and how they fit with the Essential and Desirable criteria reported below.
Selected students will be invited to an interview.
- To be able to proficiently interview people in Dutch and in English
- Ability to meet the deadlines
- Availability to travel
*to apply for this project you need to have at least the essential criteria
- Previous experience in interview methods and in applied human factors methods
- Previous experience in qualitative and quantitative methods of data gathering and analysis
- Previous experience in literature analysis
- Proficient ability to present in PPT
3. The main objectives of this project are to:
- explore current clinical pathway for people with mental and neurological disorder, and prioritise one pathway for your research;
- define the problem and your hypotheses, and develop an evaluation protocol (in agreement with Philips)
- identify health service performances variables (time, people involved, number of tasks, likelihood of errors etc.), and economics constraints to model benefits and disadvantages of introducing NeuroAI in the selected clinical pathway.
- Generate evidence to design prototypes of future/enhanced pathways;
- Evaluate these enhance pathway with experts. Evaluation methods could also include the use of eye trackers to establish the value of the new pathway. This will be decided in tune with the experimental hypothesis.
- Evaluate trust toward NeuroAI and acceptance, as well as potential barriers of adoption, with questionnaires or techniques to be decided.
4. Methods for research
The student will apply qualitative and quantitative methods, such as task analysis, interview, questionnaires and focus groups. Ad hoc methods will be selected in tune with emerging needs and in tune with problem structuring and hypotheses. Data analysis will be applied including thematic analysis, descriptive and inferential statistics.
5. Procedure of internship and thesis
- Exploratory phase (Internship, 10 EC): Two months of internship at Philips atDepartment of Digital Platform Solutions, Philips Research, High Tech Campus 34, 5656 AE Eindhoven. This experience will enable you to perform a domain analysis and a phase of problem structuring. During this phase you will identify hypotheses, and prepare an evaluation protocol to perform a small scale evaluation. A final report will be produced and presented at the end of this phase exploratory phase.
- Thesis project (25 EC): You will use the results of your internship by focusing on specific issues and variables emerged from your exploratory phase. To define (extend) hypotheses and to design evaluation protocol and experiments
6. Project starting date:
As soon as a satisfactory candidate will be identified throughout an interview the project will be assigned. Expected start date: Mid-March/April 2018.
7. Expectations and duty:
- mandatory weekly report of activities and meetings with the internal supervisor
- Prepare and manage interview with professionals around the country and arranged in agreement (and with the support) of Philips;
- Data management and storage will be student responsibility. Cloud shared folder will be organized with supervisor.
- Deadlines will be agreed among student and supervisors in tune with the planned activities.
Report of internship project June/July 2018
Master thesis and PPT in Jan 2019. Preliminary version of data and material will be also shared with Philips.