| 摘要 |
Enhanced realism of a real-time simulator having multiple computer-controlled units results from making the units capable of reacting to only those other units that each of them can be aware of because of their spatial relationships to the unit. Awareness is based upon probabilities; it can persist after a relationship changes; and it can be influenced by a units designation of a target. Each unit selects a target based upon a score incorporating multiple aspects of its tactical situation, and can change targets when the situation changes. A unit selects a strategy in response to which of a set of tactical configurations exist between the unit and its target; the strategy can change short of completion when the configuration changes. A plan produces guidance commands from the high-level strategy. The guidance commands are converted into control settings for guiding the subject unit using a physics engine for simulating the physical dynamics of the unit. The control settings can interact with each other under certain conditions. The time rate at which each computer-controlled unit performs the above operations varies with the tactical situation, and can be different for different units. The operator for each unit precomputes some data required for multiple calculations in performing the operations.
|
