The fields of neuroscience and psychology offer control engineers plenty of inspiration for evolving simple controllers into controllers that are more sophisticated, optimized, adapted, and intelligent. Typically, the control tradeoff exists between optimality and robustness. The goal of adaptability is also highly desirable. Adaptive controllers feature adjustable parameters and update mechanisms that respond to variations and disturbances. Neurocontrollers take a different approach. They are based on black-box modeling (and sometimes hybrid approaches). Such controllers function quite well in structured environments but not so well in dynamic, uncertain environments, leading to the necessity for human intervention. Here is where psychology comes into play. If the controller could collect the sufficient information it needs to perform tasks and organize itself, human operators would not be necessary in the control loop, and the means to do that is to build cognitive functionality into the system.
展开▼
