Fault Detection in Lateral Vehicle Control. Simulation of an automobile under automatic steering control, from the occupant’s point of view Evaluation of an occupant’s ability to detect and correct for controller malfunctions that will cause the vehicle to drift out of its lane
The human controller brings certain properties as yet
unduplicated in machine controllers. These include:
1) Superior image processing
2) Superior adaptability based upon cognitive skills
3) Superior ability to anticipate
But slower responses, poorer control performance, divided attention, fatigability, diversity of skills, and distractibility are distinctly human traits as well. What, then, is the right strategy to combine human and machine fault detection capabilities?
We view the range of possible roles as lying on a continuum.
At one end (H), we have the human fully in charge, at the other end,
the controller (C) is in charge. We have focused our attention on
three intermediate points or cooperative strategies (CS), between
1) as helper 2) as peer 3) as crisis handler
We have developed a Scenario Evaluation System (SES) in which
to explore both the fault detection and the fault handling
performance of each of these three scenarios.
The SES is essentially a pre-simulator. It is a system in which we
isolate many variables to concentrate on just a few. Our main
interest is the role of vision in the performance of the combination
formed by the human operator and the controller. We thus exclude
important variables from consideration. The advantage is
specificity concerning the role of vision. Disadvantages are
overcome by migrating the study to either a real simulator or
to the test track.