The supervisory control of unmanned vehicles is likely to be an important form of next-generation human-machine interaction. Although the effective design of control interface is critical for high-performance human-robot teams, there is little framework beyond general design principles in the Human Factors discipline. The main challenge is to find an optimal balance between knowledge-driven control functions and intuitive maneuvers. In general, the supervisory control of unmanned vehicles requires its operator to map the vehicle’s motion parameters with the set of control functions implemented in an interface. Due to the complexity of the control functions and underlying domain-specific knowledge, it usually takes significant time and efforts to learn the mapping relationship and familiarize oneself with the interface. In this regard, intuitive control interface is an obvious virtue that can save the cost of learning the interface, as well as acceptance by a larger group of users. With increasing types and numbers of unmanned vehicles/robots, a lack of intuitiveness can bring about substantial usability issues, including the cost of learning how to control a new vehicle, and the cost of switching to different types of vehicles. Despite the needs, the notion of intuitive control has little theoretical foundation, thus, difficult to implement through design practices. It is the ultimate goal of the current research to generate design principles that balance between knowledge-driven control and intuitive control by establishing an analytic framework of cognitive task monitoring. The analytic framework intends to estimate the cognitive processing underlying a sequence of control actions, thereby, provides empirical evidence of intuitiveness versus knowledge-dependency in control. The current research uses a Bio-inspired Underwater Vehicles (BUV) to apply the analytic framework under a variety of operational scenarios to monitor the operator interaction. To evaluate the degree of intuitiveness versus knowledge-dependency, the existent interface built in LabVIEW (Ver. 2017, National Instruments, Corp., Austin, TX) is being tested on a group of experts and novices under a variety of task scenarios. As a result, the current interface is evaluated regarding the cost of learning, i.e. the degree of reliance on knowledge, and the cost of switching to different control functions, i.e. the degree of counter-intuitiveness. Finally, the analytic outcomes lead to the redesign of the interface.
