A framework for knowledge-based team training
MetadataShow full item record
Teamwork is crucial to many disciplines, from activities such as organized sports to economic and military organizations. Team training is difficult and as yet there are few automated tools to assist in the training task. As with the training of individuals, effective training depends upon practice and proper training protocols. In this research, we defined a team training framework for constructing team training systems in domains involving command and control teams. This team training framework provides an underlying model of teamwork and programming interfaces to provide services that ease the construction of team training systems. Also, the framework enables experimentation with training protocols and coaching to be conducted more readily, as team training systems incorporating new protocols or coaching capabilities can be more easily built. For this framework (called CAST-ITT) we developed an underlying intelligent agent architecture known as CAST (Collaborative Agents Simulating Teamwork). CAST provides the underlying model of teamwork and agents to simulate virtual team members. CAST-ITT (Intelligent Team Trainer) uses CAST to also monitor trainees, and support performance assessment and coaching for the purposes of evaluating the performance of a trainee as a member of a team. CAST includes a language for describing teamwork called MALLET (Multi-Agent Logic Language for Encoding Teamwork). MALLET allows us to codify the behaviors of team members (both as virtual agents and as trainees) for use by CAST. In demonstrating CAST-ITT through an implemented team training system called TWP-DDD we have shown that a team training system can be built that uses the framework (CAST-ITT) and has good performance and can be used for achieving real world training objectives.
Miller, Michael Scott (2006). A framework for knowledge-based team training. Doctoral dissertation, Texas A&M University. Available electronically from