Seeing with Space: Applications in Computer Vision & User Feedback

Seeing with Space: Excerpt from a Literature Review

Various factors need to be considered when evaluating the utility of visual interfaces for preserving a driver’s situational awareness in a vehicle. When investigating the decision-making processes of a driver, researchers have scrutinized the many ways in which interfaces present content. Continuous spatiotemporal feedback may be able to guide a driver’s attention more reliably than a display-by-exception approach. With continuous feedback, drivers can form a more complete conceptual model about risk factors. The following section describes the recent efforts that have been taken in understanding the different uses that visual, auditory, and tactile warnings have had in alerting the driver of the spatiotemporal relations of objects in their internal and external car environments.

Visual Warnings

When interpreting a specific system failure, users only need to understand essential information about an event; users only want to see displays that support pertinent goals (Palmer & Abbott, 1994). This is known as a display-by-exception approach. However, a display-by-exception approach to interface development should be supplemented with continuous feedback when developing an interface for any fast-moving vehicles. With this in mind, driving researchers Seppelt and Lee (2007) assessed whether a continuous display supported a driver’s ability to transition from a state of adaptive cruise control to manual control when the car's sensors started to degrade from rain droplets. Continuous interface design can provide drivers with critical information about the spatiotemporal properties of nearby cars, such as time to collision, time headway, and relative velocity between vehicles (see Fig. 1). Drivers using the head-down continuous display in Seppelt and Lee’s study were required to follow a lead car that braked abruptly. They exhibited quicker braking reaction times and greater (and thus safer) headway profiles than drivers using display-by-exception interfaces. As per the requirements of a secondary task, drivers also tended to respond to auditory questions about nearby restaurants more quickly when using the continuous display. This continuous and spatially coherent interface reduced the need for the average driver to directly monitor their external environment. Drivers’ conceptual models of the car's intended actions, along with events happening on the road, were reinforced. The researchers recommended the application of continuous spatial displays for faster and more accurate primary and secondary task performance.

In support of continuous feedback, Stanton et al. (2011) evaluated the effect that a spatially informative interface had on a driver’s successes in identifying relevant road targets (see Fig. 2). Their prototypical interface, located in the car’s instrument cluster, not only communicated information about the range between the host and target cars, but also signified the presence of a new target car via a salient flashing icon. The task required drivers, who were using Stop & Go adaptive cruise control, to track the onset and offset of lead target cars in their environment and report their own vehicle’s mechanical behaviours. Drivers using this spatial interface were better at identifying the onset of a new target compared to drivers using the non-spatial flashing icon interface. This was because the spatial interface allowed drivers to perceive the provided range information and differentiate the introduction of every new target from the absence of every old one in a multiple target tracking task. Although driver situational awareness was enhanced, post-hoc comparisons revealed that the subjective workload increased with the more complex spatial display. Nevertheless, drivers actually preferred this display due the comprehensive and continuous information it provided in ambiguous task scenarios. It seems that relevant spatial feedback may help a driver discern what is happening on the road, especially in complex multi-target settings. But, emphasis should also be placed on ensuring that cognitive resources are minimally used in the process. This is especially important when a driver is distracted by a secondary task and must suddenly transition to an alerted state.

References

Seppelt, B. D., & Lee, J. D. (2007). Making adaptive cruise control (ACC) limits visible. International Journal of Human-Computer Studies, 65(3), 192–205.

Stanton, N. A., Dunoyer, A., & Leatherland, A. (2011). Detection of new in-path targets by drivers using Stop & Go Adaptive Cruise Control. Applied Ergonomics, 42(4), 592–601.