查看文章

An incremental process for developing human-agent-robot applications uses mixed-reality testbeds of varying fidelity and size. to develop such applications. Because of the requirements of efficiency, reliability, and robustness of such systems in real-world conditions, no development methodology will be effective without strong support for realistic evaluation and testing. In contrast to standalone software systems, the operation of HART applications depends on factors beyond the actual software logic—in particular, on the characteristics of the hardware (sensors, actuators, and communication links), the dynamics of the environment, and the behavior of the humans involved. A reliable assessment requires a testbed that approximates these factors with a sufficient level of fidelity. In general, testing an application in the full target configuration (that is, with the complete set of hardware assets and human individuals operating within the target physical environment) provides the most reliable assessment. Unfortunately, such fullconfiguration tests are expensive in terms of money, time, and resources, and could carry substantial risks—for example, testing a collision avoidance functionality between unmanned aerial vehicles (UAVs). This makes full-configuration testing impractical in the early stages of application development, when developers must perform a lot of evaluation and testing quickly to assess multiple design options. To reduce costs and risks, application developers can employ simplified testbeds that approximate the target application setup. These testbeds can fully or partially replace the environment, hardware, and human actors with …