In relation with the increased performance of computers, especially concerning 3D graphics, progress is achieved very rapidly in the field of Virtual Reality. As there are now already PC’s with high performance graphic boards available at reasonable prices, interactive 3D applications become increasingly attractive to different application areas. However, progress made concerning the development of highly interactive synthetic environments is not very large. Commercially available VR-Systems like World Toolkit from Sense8, DVise from Division, Coryphaeus, Superscape from VRT, Realax from Media Systems, etc. provide comprehensive sets of functionality but they are multi-purpose applications. Considering the demands of AITRAM the drawbacks of these systems are more important than the advantage, that they already provide some of the functionality needed.

Fraunhofer IFF has at its disposal concepts and different modules which were especially developed for training of maintenance technicians. Therefore the consortium will make use of these existing solutions. AITRAM does not start from scratch but with the best choice between fit for purpose and flexibility of own development. In general, one can determine that existing VR applications provide only very limited interactivity. AITRAM will have to improve this and let the trainee be a real active user. Most system’s architectures and data structures provide only poor support for training applications. Additionally, all companies providing VR systems can’t be seen as already established on the market, except perhaps Division Ltd. Therefore, there is a high risk to rely on the specific formats used by these systems. Especially the objective of interfacing or integrating the Virtual Environment with the simulation of a work environment and virtual colleagues requires flexibility in terms of implementation and insight in the internals of the simulation loop of the Virtual Reality application. Therefore, the consortium prefers to base the development on previous work done by partners. The underlying tool-set includes Performer from SGI and C/C++. Existing projects which are placed in similar domains, are ASSIST and TRAIMWE, both funded under ESPRIT4. None of them integrates HF aspects into the training concept in the way AITRAM does.

Existing solutions in CBT

In general, it can be determined that many companies already use CBT also for maintenance training. However, this refers only to acquiring theoretical knowledge, not to practice/capabilities. These CBT applications make use of the general state-of-the-art concerning multimedia. Very advanced solutions make use of VRML1/2. This is a very promising way, but the mentioned approaches have all in common several drawbacks. Also for the most advanced VRML approach these are: very limited interaction of the user only indirect interaction with the training object or parts of it no immersion of the user, no natural user interface no stereoscopic vision no tracking of the user (neither head nor hand) no high-fidelity visualisation.

Human Factors

Human Factors Aircraft maintenance is a training-oriented industry. General and specific competencies are developed through basic and type training. These competencies are enhanced and updated through continuation training. If a particular inadequacy is identified, a new type of aircraft is purchased, or new maintenance techniques have been developed, appropriate courses are sought out for selected technicians. As Human Factors have emerged as critical to safety of maintenance, a major response by the industry has been to introduce Human Factors Training. The natural response is “if Human Factors are such a problem, then let’s train our technicians in Human Factors”. Human factors training has been developed by airlines, manufacturers and training organisations; it has now been mandated by the JAA and FAA. Much of the existing training in the industry consists of discrete one- or two-day courses for maintenance personnel aimed at raising their awareness and knowledge of Human Factors.

The initial challenge of Human Factors is to change the existing culture of the company from one which is primarily technically oriented to one which is equally competent at managing the human aspects of the operation. The challenge for the future will be to sustain and enhance this new culture. A key element of this is training of personnel entering the company. Initial training of technicians is explicitly about teaching them the knowledge and skills they will need to do their jobs. But there is another, more implicit, type of learning occurring. Technicians are learning the professional culture of aircraft maintenance. If Human Factors is treated merely as another subject on the curriculum, a great opportunity will be missed to influence the formation of this professional culture in the next generation of technical personnel. In conjunction with Dublin Institute of Technology (DIT), TCD has been exploring the potential for integration of Human Factors into the technical curriculum. DIT provides basic training for most of the technicians in Irish industry. Integration of technical and Human Factors Training comprises the incorporation of Human Factors material into the instructional content of technical classes. With Virtual Reality and Simulation the integration of technical and Human Factors Training can be achieved in an efficient way.

Learning Methodology

- Didactic Besides technical aspects of innovation, there are others e.g. in the area of instructional science. The merits of customised or tailored training for individual trainees with different learning styles and experiences have been realised since the early 1980s. Traditional computer-based training systems have relied on behavioural psychology emphasising the analysis of jobs into a set of core skills. Components of skills could be mastered by following a prescribed and rigid sequence of exercises. The lack of opportunities to adapt the training exercises to the different learning needs of trainees and difficulty in improving the theoretical knowledge of trainees have led to further developments.

A review of the literature on training e. g. troubleshooting skills [5] has pointed out the need for developing computer-based training systems that could employ ”learner models” in order to customise interactions to different stages in mastering skills for different individuals. In the late 80s we have witnessed the emergence of ”intelligent” tutoring systems (ITS) which used Artificial Intelligence in order to create models of the learner and the expert to provide customised instruction. Reviews of ITS have been provided by various books [2], [6]. Common criticism of many ITS include: the large amount of effort in preparing the training system, the difficulty of the human instructor to co-ordinate with the system author, the simplicity of application domains, and mainly the lack of empirical data to demonstrate significant differences with more traditional forms of computer-based training. Most of these training systems have been developed in the context of troubleshooting and repairing aviation equipment in the military sector such as navigation equipment, radars and aircraft engines.

One of the main factors of success of these training systems was the teaching of ”mental models”, that is, practical theoretical knowledge which could enable trainees to transfer or generalise their skills from one situation to another. Earlier work in the area of qualitative or mental models of equipment had provided a useful basis to explore the issue of transfer of skills. The increasing advances in new technology and electronics have presented maintenance technicians with the problem of catching up with many variations in the equipment they had to maintain. The issue of transfer of skills, therefore, is very important for responding to the need of the industry to upgrade to ever increasing sophisticated machinery. Another lesson learned from the evaluation of ITS is the need to employ simulated models of devices or machinery in order to facilitate instructional strategies such as learning by doing and guided discovery. Simulated models of equipment offer excellent opportunities for acquiring skills such as ”how-the-system-works” and ”how-to-do-the-job”. A major advantage of simulation is that trainees are able not only to acquire a skill quickly but also to retain their skills in the long run.

This leads us to consider three important criteria for developing and evaluating new computer-based training systems. That is, acquisition of skills, transfer of skills to new but similar equipment and retention of skills even when practice on the job is not frequent. Virtual reality (VR) is a promising area for further developments in simulated models of devices and consequently computer-based training. With VR it is possible for the trainees to gain access to more realistic representations of the equipment they are likely to encounter in their daily jobs. It is anticipated that VR will enhance human memory of system components and required skills and thus, knowledge will better retained and transferred. However, applications of VR to computer-based training are lacking and there is a growing need for investing in such research developments. Our proposed system (AITRAM) will integrate VR with various pedagogic issues such as training modes and competence measures. Specifically, we would like to explore how AITRAM can provide a variety of training modes, which include:

Learning by doing

• where the trainee masters a skill by actively troubleshooting and repairing devices simulated in the AITRAM environment
• guided discovery or ”scaffolding” where the trainee undertakes a complete exercise even from the early stages of learning with the support of the system.
• part-task training, that is, mastering component skills and then practising them as a whole
• teaching qualitative models or mental models of the simulated device
• heuristic training, where the trainee masters a common body of heuristic rules that help him narrow down possible device faults.

We will provide AITRAM with a variety of measures for assessing different aspects of competence. Specifically, AITRAM would address the following measures of competence or achievement:

• average time to troubleshoot or repair
• accuracy in troubleshooting, i.e. how many faults have been identified in a set of exercises referring to the documented procedure
• economy of strategy which reflects the number of information sources consulted in order to identify a device fault. Redundant checks, for instance, could indicate troubleshooting strategies which take more time to implement.
• errors of omission and commission in exercises which present trainees with a ”grey” fault symptom compatible with more than one faults.
• recovery of errors committed earlier on in the learning sequence. This is an important aspect of competence in complex devices where errors in troubleshooting are bound to occur. Detecting and recovering errors as much important as preventing errors in the first instance. Methods such as error training can be used to facilitate error recovery [9], [10].
• improvements in mental models of trainees which may not show up in the series of tests or exercises to be undertaken by trainees.

homepage