In cooperative training, humans within a team coordinate on complex tasks, building mental models of their teammates and learning to adapt to teammates’ actions in real-time. To reduce the often prohibitive scheduling constraints associated with cooperative training, this article introduces a paradigm for cooperative asynchronous training of human teams in which trainees practice coordination with autonomous teammates rather than humans. We introduce a novel experimental design for evaluating autonomous teammates for use as training partners in cooperative training. We apply this design to a human-subjects experiment where humans are trained with either another human or an autonomous teammate and are evaluated with a new human subject in a new, partially observable, cooperative game developed for this study. Importantly, we employ an unsupervised sequential clustering methodology to partition teammate trajectories from demonstrations performed in the experiment to form a smaller number of training conditions. This results in a simpler experiment design, enabling us to conduct a complex cooperative training human-subjects study in a reasonable amount of time. Through a demonstration of the proposed experimental design, we provide takeaways and design recommendations for future research in the development of cooperative asynchronous training systems utilizing robot surrogates for human teammates.

A technique is presented for obtaining a specification from a requirement through a series of incremental steps. The starting point is a problem frame description involving a requirement on the phenomena of the problem domain, and a decomposition of the environment into domains, connected to one another and to the machine being implemented by shared phenomena. In each step, the requirement is moved towards the machine, leaving behind a trail of `breadcrumbs' in the form of domain assumptions. Eventually, the transformed requirement references only phenomena at the interface of the machine and can therefore serve as a specification. Each step is justified by an implication that can be mechanically checked, ensuring that, if the machine obeys the derived specification and the domain assumptions are valid, the requirement will hold. The technique is applied to the logging subproblem of a radiotherapy system

A method for extracting syntactic specifications from heapmanipulating code is described. The state of the heap is represented as an environment mapping each variable or field to a relational expression. A procedure is executed symbolically, obtaining an environment for the post-state that gives the value of each variable and field in terms of the values of variables and fields of the pre-state. Approximation is introduced by forming relational unions at merge points in the control flow graph, and by widening union-of-join expressions to transitive closures. The resulting analysis is linear in the length of the code and the number of fields, but capable of producing non-trivial specifications of surprising accuracy.

Two operations commute if executing them serially in either order results in the same change of state. In a system in which commands may be issued simultaneously by different users, lack of commutativity can result in unpredictable behaviour, even if the commands are serialized, because one user's command may be preempted by another's, and thus executed in an unanticipated state. This paper describes an automated approach to analyzing commutativity. The operations are expressed as constraints in a declarative modelling language such as Alloy, and a constraint solver is used to find violating scenarios. A case study application to the beam scheduling component of a proton therapy machine (originally specified in OCL) revealed several violations of commutativity in which requests from medical technicians in treatment rooms could conflict with the actions of a beam operator in a master control room. Some of the issues involved in automating the analysis for OCL itself are also discussed.

Declarative models, in which conjunction and negation are freely used, are susceptible to unintentional overconstraint. Core extraction is a new analysis that mitigates this problem in the context of a checker based on reduction to SAT (systems analysis tools). It exploits a recently developed facility of SAT solvers that provides an "unsatisfiable core" of an unsatisfiable set of clauses, often much smaller than the clause set as a whole. The unsatisfiable core is mapped back into the syntax of the original model, showing the user fragments of the model found to be irrelevant. This information can be a great help in discovering and localizing overconstraint, and in some cases pinpoints it immediately. The construction of the mapping is given for a generalized modeling language, along with a justification of the soundness of the claim that the marked portions of the model are irrelevant. Experiences in applying core extraction to a variety of existing models are discussed.