Controlling Steering Using Vision

The Two-level model is a popular account of how humans use visual information to successfully control steering within road edges. A guidance component uses information from far regions to preview upcoming steering requirements, and a compensatory component uses information from near regions to stabilise position-in-lane. Researchers who have considered the case of driving often treat road edges as the sole informational input for controlling steering, but this approach is not consistent with the notion that the human visual system adaptively uses multiple inputs to maintain robust control of steering. A rich source of information which may also be useful for steering control is optic flow. Chapter 2 demonstrates that optic flow speed is used to control steering even with road edges present. Chapters 3-5 develop a framework to examine how use of flow speed changes depending on the availability of guidance or compensatory road edge information, and demonstrate that use of flow speed increases only when guidance level information (far road edges) is present. Chapters 6-7 go on to examine the contribution of flow direction to controlling steering within road edges, and demonstrate that the use of flow direction appears to be yoked to the presence of compensatory information (near road edges). Together, these experiments demonstrate that the contribution of flow information to controlling steering within road edges can be understood within the context of two-level steering, and show that an approach which emphasise robust control through combining multiple informational inputs is vital if we are to fully understand how the visual-motor system solves the problem of steering along constrained paths