Abstract
Many research projects have identified three major obstacles to a broad implementation of Design for All: lack of awareness among users, designers and suppliers, technical feasibility and commercial viability. Mainstream manufactures do not have a detailed understanding of the needs of people with disabilities. This paper presents an approach to use standards-based Virtual User Models that covers mild and moderate disabilities to support designers in understanding these needs. This approach consists of a virtual laboratory with three design phases to allow designers to plan and evaluate the user interfaces of their products. We review here the state of the art and present our Virtual User Model as a mixture of human and environment context.
Keywords: virtual user model, computer design, design for all, accessibility, usability, ontology.
Introduction
This paper presents the output of a study of existing virtual user models and virtual user model-based approaches in combination with CAD design tools under the scope of the VICON research project.1 Additionally, we describe the development of a methodological framework of an evaluation scheme for a virtual user laboratory.
Understanding and incorporating the requirements of persons suffering from physical impairments will only be achieved when designers are able to see more clearly the impact of specifications and design decisions, regarding the accessibility of their envisaged products. Simulation of the causes and levels of design exclusion using virtual users has the potential to provide such insight.
We will review:
— existing approaches in the area of human factors in the workplace and product design;
— task analysis equations and tools identifying design problems to avoid the causes of less usable or inaccessible designs; and
— additional analysis types and extensions to demonstrate the VICON objectives in relation to designing mobile phones and washing machines.
The envisaged beneficiary user group in VICON is represented by users with mild to moderate physical impairments. The intention behind the project is to come up with a design support tool in order to highlight areas of particular accessibility issues during the design process of a product before the realization of any hardware prototype. This approach incorporates a user capability range that does not cover the "average user", thus takes into account multiple combinations of disabilities.
The focus in VICON lies on the user interface of the product, and subsequently touches the issue of interaction of persons with physical impairments with consumer products. This plays a crucial role in determining how well a particular consumer product is accessible and suitable for use by elderly and disabled persons. Generally, how a human will behave and interact in relation to a product or system is particularly difficult to predict. Yet, usability or accessibility issues traditionally have been often addressed by intuition. Usually, real tests are performed after the product can be easily modified. Very often this results in the decision by the design teams of neglecting the accessibility demands on such products.
Designers therefore need support by simulating the interface between users and a consumer product from the earliest stages of the design and engineering processes. Evaluating alternatives from an accessibility standpoint, when it is still relative inexpensive to change the design, can improve the accessibility features of the product [1]. VICON strives to provide such a support from the initial stages until the beginning of the prototyping and production phases.
The approach pursued in VICON has the potential of providing an in-depth understanding of crucial problems encountered by design for all, including elderly and disabled persons. This will support designers to incorporate the requirements of users with disabilities, without putting the burden on the designers to know a lot about the different disabilities and their grades and combinations, and how to compensate them by applying specific design patterns [1, 4].
Looking at contemporary virtual user models, a lot of existing approaches were identified -ranging from figures used to create cartoons and simple games to avatars used in Second Life and alike. Only the approach of the digital human models based on anthropometrical data are suitable for serious simulations and can meet the ambitious goals of the accessibility evaluation of user interfaces of consumer products.
There is neither a common framework, nor a common understanding of how elements involved in the creation, development, and interaction of virtual user features are done [1]. Therefore, there is a need for describing (1) existing resources, (2) virtual users composition and features, and (3) the different levels/fields of knowledge comprehended.
Investigations of existing systems, approaches and models were based on literature review and upon hands-on experiments with some virtual user systems, accessible as public versions or as licensed agreements with the vendors. Another source of information was based upon currently ongoing or finished EU funded research projects in the same area, such as the VAALID2 and the AEGIS3 projects. The main differences of existing projects compared to VICON were (1) the direct user target group, which comprehends here designers and not the beneficiaries themselves, and (2) the focus on specific target devices, which are in VICON washing machines and mobile phones. Available experiences with the inclusion of virtual users have been analyzed in order to base the VICON approach on a foundation of past experiences, and derive an improved methodology. Starting with a thorough study of literature of virtual user approaches, this will infer requirements and best-practices. The virtual users will be modeled in a hierarchical way, one level describing a target user group in a more general way, the next level modeling instantiations as "virtual users" explicitly. This will also serve traditional methods, such as scenario-based analysis, while allowing comparing them with Virtual Reality-based methods. Benefits of such virtual user models strongly depend on two qualities: (1) the validity of the user model, i.e., the match between model and reality, and (2) its capability to predict the user behavior correctly. Further, we considered the flexibility of the virtual user models, i.e., their capability to apply the model during all phases of the design process [4].
State of the Art
A virtual user concept is commonly utilized for ergonomic analysis in vehicle and workplace design in the automotive industry. It is used to validate the design in a simulation environment, check in a loop if the design is suitable, refine it considering recommendations and best practices and finally, when found suitable, produce a prototype to be checked by end users (see Fig. 1). More rarely are the tools applied in the evaluation of consumer product designs and even less for usability and accessibility of consumer product’s interfaces, although having similar objectives for user-centered design processes.
In the area of accessibility only one case-study was identified, which is the HADRIAN system based on the SAMMIE CAD [5], which tried to detect accessibility issues during the interaction between users and ATM machines. Virtual user systems for the purpose of validation of product and workplace design are based upon anthropometry, joints’ range of motion, description and appearance of the virtual user customized to meet the requirements of the task at hand. Virtual user models (digital human models) are as mentioned above already established tools in many companies for specific analysis tasks in product design or in process design and development. In this survey we have gathered information about the state of the art of models used in CAD (Computer Aided Design) and PLM (Product Lifecycle Management) systems [6].
The majority of existing models are mainly developed for the areas of aerospace and design of airplanes, where training and experiments on real tasks were expensive or impossible; therefore they had to be simulated and one element of such simulations were Virtual User Models [7].
Fig. 1. General usage approach of virtual user models in inclusive design 2.1 Characteristics of Existing Digital Human Models
The main characteristics of existing virtual user models include a definition of the physical environment, a definition of virtual users segregated into segments and joints or links, and that the virtual users should perform the analysis with a reasonable posture. Furthermore, a generic process for virtual user model analysis includes the following major steps: understanding the task, understanding the work environment, understanding the user population, understanding the limits of the software used, performing the analysis, analyzing and applying judgments to the results, and finally, reporting the results of the analysis. The documentation of analysis results also needs to be structured as a natural part of the virtual user modeling process. In the following we present the main components of a virtual user system in existing systems [3]:
— Virtual user. It represents the humans who are interacting with the product or vehicle. The existing virtual user models represent the human body as a kinematic system, a series of links connected by rotational degrees of freedom (DoF) that collectively represent musculoskeletal joints. Standard set-ups are usually made available, so that the designer can select from a list of existing stereotypes of virtual users such as: customer type, power plant worker type and manager type. It is also possible to specify a unique type of virtual users, which is then generated in the human simulation tool by defining the virtual user characteristics, e.g.: nationality, age, gender, percent of target population considered and key anthropometrical variables.
— Physical environment. It refers to descriptions of the workplace, the device characteristics or vehicle parts which a human interacts with by performing tasks. The environment is described in the detail that the analysis requires. Relevant information is applied, such as size in a clearance analysis, and weight in a force/torque analysis. The numbers of the drawings used to create the simulated environment are stored. In addition, simplifications as well as limitations of the environment descriptions are explained.
— Tasks. The task is the action that the virtual user will perform. Initially, the task is divided into subtasks using hierarchical task analysis in order to retrieve simple tasks that can be handled and simulated. Secondly, constraints for performing the task are defined. Standard constraints for different tasks are available and visualized in the process with an illustration of a driver with marked constraints.
— Results. In the results, it is possible to attach generated animations, pictures and tables of the analysis to the documentation. It is also possible to write text with illustrations or to just describe results in text.
Existing Virtual User Models for Ergonomic Analysis
There are many Virtual User Model platforms and products available mainly in the area of product lifecycle management like, e.g., Tecnomatix (Siemens/UGS) which is using a model called eM-Human. The model Santos was developed in the frame of the Virtual Soldier Research Program of the University of Iowa. It uses accurate biomechanics with models of muscles, deformable skin and the simulation of vital signs. With this system analyses of fatigue, discomfort, force or strength can be done. Furthermore, modules for clothing simulation, artificial intelligence and virtual reality integration are available for real-time systems. Unfortunately, this model is not available for the research community in Europe. Some other models like the Boeing Human Modeling System (BHMS) or the System for Aiding Man-Machine Interaction Evaluation (SAMMIE) complete this listing. Many problems can be solved with nowadays Virtual User Models. Nevertheless, there are still many unsolved issues which can be developed and integrated in the future, by enhancement of existing or development of in new models.
The VICON Approach
The VICON evaluation methodology consists of many components (see Fig. 2. and Fig. 3) and is based on existing approaches of virtual user systems in the areas of ergonomic and usability as described in section 2.2. The VICON approach goes beyond the existing approaches by:
— accompanying the design process from the scratch until the final CAD design cycle, providing different standalone recommendation systems for the idea finding stage, the scratch phase and components for integration with CAD systems based on the VICON virtual user concept;
— incorporating accessibility features in the user profiles, especially audio analysis, which was not considered in any of the existing virtual user platforms;
— developing further analysis tools to assess accessibility requirements in the virtual user system; and
— adding new reporting mechanisms, e.g., web-based reports and visual responses in the virtual user itself.
A recommendation system is being created to assist the designer at the first stages of ideas finding and scratch design. The recommendation system will be made available as a standalone application and will provide an API to be integrated into other applications.
Following an example is given of the VICON recommendation system as a web based virtual user recommendation system on inclusive design for mobile phones and washing machines to assist designers from the beginning utilizing existing guidelines, standards and materials.
Fig. 2. Overview of the VICON process
Fig. 3. Overview of the VICON approach
The designer is planning to design a new mobile phone and wants to incorporate accessibility features into the future mobile phone. The designer invokes the standalone VICON recommendation system and selects from the device list "mobile phone" and configures from the "target user group" list its target user group. Based on the entered information the system looks-up into its repositories and displays to her a list of existing use cases as defined in earlier stages of the VICON project and additionally links to existing guidelines. In the further course the designer starts a graphics application and makes a scratch layout of the future mobile phone, saves it and imports it to the VI-CON recommendation system. The system then displays more accurate recommendations dedicated to the scratch layout and the selected target user group.
The other component will be the 3D virtual user system, which allows either the development of CAD designs or the import of CAD designs into it, and provides analysis tools. It will allow the assignment of tasks to the virtual user. An execution procedure reports the results visually to the designer.
The user of the VICON virtual user model executes the following steps, which comprises the virtual user system:
1. Define parameters, which enable the consideration of dynamic behavior and interaction between the components of the VICON system,
2. Incorporate task analysis tools, which formalize and structure the performance of a virtual user in a sequence of goals and actions carried out during the interaction,
3. Create virtual user environments, where the designers could configure a virtual environment,
4. Design products in a CAD system, where the designer can invoke a VICON context sensitive recommendation system,
5. Build a virtual environment,
6. Import the CAD design into the VICON virtual user system,
7. Position the virtual user in the environment,
8. Assign tasks to the virtual user, and
9. Analyze how the human performs.
4 Criteria and Methods to Validate the Vicon Model
The objective of evaluation of the VICON virtual user system is to examine the validity of such models and to what extent accessibility simulations of interaction tasks with washing machines and mobile phones designs correctly predict the real outcomes of the accessibility evaluation by real users using real devices. Furthermore, it checks whether recommended guidelines originating from accessibility simulations are correctly considered. The evaluation was conducted in a virtual lab setting by the industrial partners of VICON. The results should show whether the virtual user models and related analysis tools are useful for providing designers easily and without extra burden with accessibility hints. When the virtual user conducts tasks, it will measure, if accessibility issues are recognized correctly. The evaluation studies identified areas that require additional development in order to improve the virtual user system itself and its ability to correctly predict the accessibility evaluation outcome.
The evaluation will include an investigation about the limits of the concept of the virtual model in comparison to real world field studies. This investigation is necessary, because of the novelty of the virtual user concept in the domain of accessible design for impaired users as target group.
The evaluation should determine rules on to what extent and detail level the virtual user should emulate real world users in order to achieve optimized designs for the selected user group.
Virtual user models are visually very impressive and they can be very useful as a validation tool (e.g., for performing sight field analysis). It is already obvious that this solution will not replace real user tests [4]. One of its major downfalls is that virtual user models can not 100% build-up the real environment and there will be shortcomings in the relations and dependencies between all the variables involved in the system. However, virtual user systems can play an excellent role in automated detection of many accessibility problems in designs of products. They can play a very useful role in the communication between designers and other stakeholders to explain and discuss accessibility issues.
In the following, we suggest criteria and methods on how to assess the VICON approach, in particular the validity of the virtual user concept.
1. Audio analysis. The analysis methods of audio issues include the configuration of an appropriate virtual environment.
Success criteria: The virtual user should display any lack of hearing sounds and signals.
2. Vision field analysis. The analysis methods of vision issues include the configuration of an appropriate virtual environment.
Success criteria: The virtual user should display any vision problem encountered.
3. Reach analysis. The analysis methods of reach issues include the configuration of an appropriate virtual environment.
Success criteria: The virtual user should display any reach problem encountered.
4. Force feedback analysis. The analysis methods of force issues include the configuration of an appropriate virtual environment.
Success criteria: The virtual user should display any force problem encountered, e.g., when pressing buttons.
5. Push/Pull analysis. The analysis methods of push/pull issues include the configuration of an appropriate virtual environment.
Success criteria: The virtual user should display any push/pull problem encountered, e.g., when opening or closing a door.
6. Lift/lower analysis. The analysis methods of lift/lower issues include the configuration of an appropriate virtual environment.
Success criteria: The virtual user should display any lift/lower problem encountered, e.g., when carrying the cloths’ basket and putting it down.
7. Grasp analysis. The analysis methods of grasp issues include the configuration of an appropriate virtual environment.
Success criteria: The virtual user should display any grasp problem encountered, e.g., when grasping the mobile phone.
8. Manipulate analysis. The analysis methods of manipulate issues include the configuration of an appropriate virtual environment.
Success criteria: The virtual user should display any manipulation problems encountered, e.g., when manipulating the tiny keys of the mobile phone.
9. Combined analysis. The analysis methods of combined issues include the configuration of an appropriate virtual environment.
Success criteria: The virtual user should display any combined problem encountered, e.g., the ability not to see clearly the buttons and not having enough strength in the fingers to press them.
Conclusions
We have investigated the existing virtual user models relevant for the analysis of Human Factors and Human Activities. We think only this type of virtual user systems would be suitable for an automated analysis of accessibility features in CAD design. We have defined the main analysis types. We are building the required prerequisites based on the use cases gathered in the virtual user model, which will be specified and built in the upcoming phases of this project. The results in the VICON virtual user system should be presented visually to the designer, so that they must not be experts on accessibility to interpret and utilize them.
In our study of existing virtual user systems we identified many possible candidates to be used in the forthcoming development of the VICON virtual user model. The first one is SAMMIE CAD, the extension HADERIAN was developed to study tasks for elderly and disabled persons. Here a database with nearly 100 users was built. The main disadvantage of SAMMIE CAD is that the hand model is not sophisticated enough to grasp and manipulate objects like mobile phones. The Pro/ ENGINEER Manikin is a highly sophisticated virtual user model based on standards like the Digital Human Model structure. It conforms to the H-ANIM standard (ISO/IEC 19774)4. However, the provided features of Pro/Engineer Manikin are more or less comparable to those i.e. of Jack, RAMSIS or CATIA human builder. The main advantage appears in conjunction that it is the preferred CAD application of the VICON industrial end-users. Hence a quick and comprehensive integration of the VICON platform into respective development process appears to be much easier.
Highly sophisticated CAD systems and virtual users based on anthropometry may be misleading concerning the validity of the models. Therefore we will test in detail all features of the VICON virtual user system to ensure its validity.
