CHAPTER 5—CONCLUSIONS
Following is a summary list of the overall conclusions in Phase III—Study 1:
- In the clear weather scenarios tested, both active and passive IR systems, if designed correctly, appear to provide pedestrian detection benefit over headlamps.
- Older people detect pedestrians later and recognize sign types later.
- IR systems may help offset the reduced visual performance of older drivers.
- Pedestrians in blue denim were detected earlier than pedestrians in black cloth for all of the VESs.
- When comparing VESs, drivers gave similar or more positive subjective evaluations to the HID headlamps, although objective measures showed similar or negative results.
- There were no differences between VESs in recognition of sign meanings.
IR vision systems provide the potential for earlier detection of objects; however, some question remains regarding the effect on the driver of when objects are not presented in a display. The following items outline IR-related findings based on the tested systems:
- The FIR and the NIR 1 provided pedestrian detection benefit in clear weather over headlamps.
- In all but one pedestrian scenario tested, detection distance was acceptable for 89 km/h (55 mi/h) driving.
- Presence of an IR system alone will not increase detection distances. Differences in implementation are important.
- Pedestrians in blue denim were detected 45 percent farther away than pedestrians in black cloth for FIR. Color differences were expected to have a minimal effect on an FIR vehicle. These differences could be explained by thicker denim cloth that may have held heat better than thinner black cloth. Also, some participants may have waited for visual confirmation (through the windshield) before “detecting” a pedestrian.
- An older group using the IR systems had similar performance to a younger group using HLB headlamps.
- Older people may have improved performance with the NIR 1 over FIR.
- IR systems can reveal pedestrians in glare earlier than headlamp-based systems.
- People with an IR system may have reduced detection distances for objects that are not shown in the system display. The FIR had the shortest detection of pedestrians on the left in a left turn scenario and the shortest detection and recognition of retroreflective objects; however, it is not clear if this was the result of headlamp performance, presence of FIR system, frequency of thermal objects in study, or system novelty and experimental situation.
- Some participants indicated they would like the display systems to show sign information.
- For most of the pedestrian scenarios tested, the FIR implementation appeared to provide a 21.3-m to 30.5-m (70-ft to 100-ft) detection advantage over the NIR 1 implementation.
- The wider FOV system (18°) tested appears to have had beneficial effects in detecting pedestrians to the side.
- The wider FOV (18°) appears sufficient for presenting pedestrians on the curve tested (radius of 1,250 m), for pedestrians located 9.45 m (31 ft) from the lane center, and for pedestrians along the sides of road. A narrower FOV (11.7°) may not present objects on curves with radii of 1,250 m or less.
From this research, development of recommendations for future direction in display-based nighttime VES systems is possible. System engineers should pursue a display that is located as close to the forward road scene as possible, using HHD- or Heads-Up Display-type (HUD) technology. Possible objects should be called out clearly in a display to minimize the need for a driver to spend time visually scanning a display. At the same time, it is not desirable for drivers to have the perception that they could drive solely by using a display. Ideally, an interface that attracts the driver’s attention when necessary but that would not otherwise require glances is preferable. This might include HUD technology, auditory warnings when a possible object is present, or activating the display only when possible objects are present. When a display is visually interrogated by the driver, the required glance time should be minimized. This might be accomplished by presenting objects in high contrast or, as enabling technology becomes feasible, by augmenting the scene with distinctive graphics to call out possible objects.
The possibility of users missing objects that were not presented on displays should be investigated further. If users are allocating attention to a display rather than to the forward road scene, what variables influence this behavior? For example, would a less realistic display increase or decrease this level of attention? Would a symbolic format or a HUD presentation eliminate this effect?
How wide an FOV is wide enough? Investigation of the performance of VESs should be undertaken to develop guidelines regarding appropriate fields of view in more severe road geometries such as on secondary roads and in neighborhoods.
Previous | Table of Contents | Appendices
