Creation of Method and Early Evaluation Package with Focus on Safety and Multimodal Evaluation for Text Input while Driving
Independent thesis Advanced level (professional degree), 20 credits / 30 HE creditsStudent thesis
A big concern in the car industry is driver distraction and inattention. Special concern is given when a task involves a user performing alphanumeric text input. These tasks often arise from interaction with the in-vehicle information system, GPS or nomadic devices such as mobile phones. To perform text input with a phone while driving increases the risk of accident by a factor 23.2 (Olson, Hanowski, Hickman and Bocanegra, 2009) and it have been estimated that 28 % drivers (of 3000) engage in text messaging activity while driving (If, 2010). Distraction and inattention problems have been reported to have caused 5474 deaths and 448 000 wounded in traffic during 2010 in USA (NHTSA, 2010a).One solution to the problems of distraction and inattention is thought to be usage of multimodal interfaces which would enhance information input and output redundancy and allow the driver to keep the eyes on the road, rather than engaging in other visually demanding tasks. This project was initiated by Semcon AB to investigate evaluation methods concerning early evaluations of early design concept of text input and was performed in conjunction with interaction design students from Chalmers University of Technology. The goal was to create a method, with multimodal variables, which would generate comparative means and help designers in the early stages of interaction design development. A literature study was conducted in order to find appropriate variables and methods for usage. The conclusion was that task decomposition methods should suffice as early evaluation as it would reveal a duration-related variable, total task time, with high correlation to other variables of special importance e. g. total visual time. However, no method was especially considering the usage of multimodality and thus a new method based on Extended Keystroke Level Model (Extended KLM) and Multimodal Critical Path Analysis (Multimodal CPA) were created. The new method was named Combined Evaluation Method (CEM) and its aim was to be very quick, flexible as and thus be able to deal with highly different concepts regarding text input and above all find relative power among design concepts.The method was iterated several times, each iteration introducing something new, and each time evaluated by experts. When the method was deemed finished (as a method that would consider most of the aspects in the requirements of text input and modal consideration) a reliability study was conducted in order to find relative power between evaluators and look for reliability in form of analysis of variance. The study could not yield reliability between evaluators, something thought to be because of participants’ lack of training in this type of method. However the results should indicate that difference between concepts in regards to total task time should be possible to estimate (a tendency towards relative power). A simple independent two tailed T test could however not prove significant, t(6)= 1.1, p>0.31.This report also contains an evaluation package, which describes noteworthy non-empirical methods in regard to text input. The package was built in order to try and synchronize different aspects of several methods to minimize application time for more extensive evaluations.
Place, publisher, year, edition, pages
2012. , 104 p.
Technology, Evaluation, safety, driving, method, KLM, CPA, multimodal
IdentifiersURN: urn:nbn:se:ltu:diva-59113Local ID: fa3c1397-bb51-4e59-badd-f6b9b41dce1eOAI: oai:DiVA.org:ltu-59113DiVA: diva2:1032501
Subject / course
Student thesis, at least 30 credits
Industrial Design Engineering, master's level
Validerat; 20120203 (anonymous)2016-10-042016-10-04Bibliographically approved