UM-HSRI-81-35 WINDSHIELD DAMAGE AND DRIVING SAFETY Paul Green William T. Burgess FINAL REPORT Tuhnicrrl Rep& Docurmtaii'bn Page 1. Roper* No. 2. Gov*mrmi Accession No. 3, Rec:ptmc's Catalog No. UM-HSRI-81-35 4. Titlo d Subtitle I= I= 5. Rlpocc Date June 1981 Wi ndshilel d Damage and Driving Safety 6. Puforainq Or~iratimCab i1 8. Perfom~mqOrqon~zmion Roporc No. Paul Green and William T. Burgess I UM-HSRI-81-35 I 1 10. WorC Unit No. (TRAIS) Highway Safety Research Institute University of Michigan 11. Contract or Grant NO. Ann Arbor, Michigan 48109 (U .$.A. ) 80-1352-C1 '3. TIP* of Report ond Period Cor*red TnTuitryN * dAddr Eommi ss ttee for Highway Safety Final Report 1717 K Street, N.N. July 1980 - June 1981 Suite 802 14. Sponrwtnq Agency Cod. Washinaton, D.C. 20036 (U.S.A.) IS. Suppl-tor). Notes I 16. Abstroct Three tasks concerned with windshield damage and driving safety were completed. First, a review of the 1 i terature revealed abundant information on general problems of visibility and the assessment of windscreen optical qua1 i ty, but 1 i ttle information specifically pertaining to windshield damage. Second, the available accident data were analyzed. While the evidence is tenuclus and sparse, there were no indications that windshield damage has been a causal factor in accidents. Only in the CPIR computerized accident file has windshield condition been coded, and in that file only two vehicles had pre-crash windshield damage. In neither case was it a causal factor. Finally, in a two-choice response time experiment, sixteen drivers viewed sl ides of nighttime road scenes through four different windshields. Drivers were asked to indicate if it was "safe to proceed" (or not) by pressing buttons. Increasing the degree of windshield damage led to increases in response time, especially when a light simulating the glare of an oncoming vehicle's headlamps was present. This last result suggests to the authors that windshield damage is a visibility problem that should receive more attention from drivers and safety organizations. 1 17. Kay Wwds I 18. Disnibutia Statmnrri highway safety, driving safety, Available from: windscreens, windshields , accidents, National Technical Information Service response time, human factors, 5285 Port Royal Road (NTIS) winds hi el d damage, ergonomi cs Springfield , VA 22161 ! 19. keuri+ Clnssii. (of *is W) IP. kwity Classif. (oi this peqd 21. No. oi Poqer 22 Pnce Nolne None 7 3 TABLE OF CONTENTS LITERATURE REVIEW. ....................... Measurement of Windscreen Optical Qua1i ty ........... Related Problems of Driver Vision, .............. Windshield Damage and Degradation. .............. ACCIDENT DATA ANALYSIS ..................... LABORATORY EXPERIMENT: NIGHTTIME DRIVER DECISION MAKING .... Method. ........................... Equipment and Material s ................. People Tested ...................... Procedure ........................ Resull ts and Discussion .................... Errors. ......................... Response Times. ..................... SUMMAR'i' AND CONCLUSIONS. .................... REFERENCE NOTE ......................... REFERENCES ........................... APPENDICES ........................... 1. 217 Cases of Windshield Condition in CPIR ........, 2, 2 Cases of Damaged Windshields .............. 3. 4.4 CPIR Cases with "Other" Visibility Code ........ 4. Test Slides: Content, Response Times, and Number of Errors 5. Instructions to Subjects ................. iii Tile relationship between windshield design and injuries from occupa~itimpact with them has been a topic of considerabl e research, both at the Highway Safety Research Institute (HSRI) (Huelke, Grabb, and Dingmati, 1964, 1966, 1967; Huelke, Grabb, Dingman, and 0 ' Neal , 1968 ; Huelke:, Grabb, and Gikas, 1966; Huelke and Sherman, 1975) and elsewhere (e.g., Patrick, 1967; Begeman, King, Weigt, and Patrick, 1978). This report focuses on a related and less well-researched question -- Do damaged windshields cause accidents by interfering with driver vision? Three topics are addressed in this section: how windscreen optical quality has been measured, general and related problems of driver vision, and studies focusing specifically on wi ndshiel d damage. Measurement of Windscreen Optical Qua1i ty The research 1 i terature offers several approaches to the evaluation of the optical quality of windshields, including driver acuity studi es , subjectiive ratings, physical measures of similarity to standards , and modu1at:ion transfer function measures. Most of these approaches are more appropriate for wide-area distortions than localized windshield damage, the interest of this report. The Chrysl er Corporation proposed a method that combines subjective ratings generated by juries of engineers with some objective, quantitative measures (Anonymous, 1963). Two properties in which Chrysler expresses interest are "daytime distortion" (due to an image a~pearingdisplaced from its true position) and "dual vision" (resulting l~hissection is an expansion of the literature review contained in the proposal for this project (Green, 1980). The previous material has been revised, and recent research findings have been added. from the internal reflection between the two surfaces of the windshield). While the optical issues are described in great detai 1 , the mechanics of the rating procedure are unclear. Kerkhof (1962) suggests that windscreen quality can be assessed by having subjects view Landholt C's, Snellen E's or some other acuity target directly through a windshield. The difference in subjects' performance in identifying targets through different wi nds h iel ds is a measure of their relative optical qual ities. A method proposed by Kohler (1973) requires photographs of a standard (reference) windshield and of a distorted (test) winds h iel d. The similarity of the two photographs is the proposed measure of optical qual ity. However, the computation suggested for this simi 1arity measure is difficult to understand. Several other studies have looked at the optical qual ities of non- automotive windscreens--in particular, those in the F-111 aircraft (Gomer and Eggleston, 1978; Ward, Defrances and Eggleston, 1979). In general, the researchers found no sing1e objective measure that completely accounted for the variance of subjective ratings. They suggested that mu1 tip1e, independent assessments were needed. Finally, a quite different approach is to examine the Modulation Transfer Function (MTF) of a windshiel d (Merritt, Newton, Sanderson an d Seltzer, 1978). Most often, the area under the MTF curve serves as a measure of the quality of the windshield. This function has been found to characterize quite accurately the qual ity of an image presented on cathode ray tubes and other types of electronic displays (Almagor, Farley and Snyder, 1978; Bi berman, 1972; Snyder, 1976). A1 so, Merr it t , Miller and Kerr (1980) found substantial re1ationships between contrast transfler values obtained with a real-time electro-optical sys tem (the "Visibility Quality Meter" (VQM)) and those reported by individual subjects across several types of windshield degradation. Merritt et al. sulggested that the VQM "stand in" for the average human eye in measurements of visibility. However,. one of the assumptions of the theory underlying the MTF approach is that the visual field is homogeneous. Once again, as a cracked or damaged windshield is extremely heterogeneous, this approach is not appropriate here. Re1 ated Problems of Driver Vision Besides methodological studies, a wide range of studies deal w i t h more substantive topics in visibility and fie1 d-of-view requirements for driver:;. Relevant topics in the visibility literature include: 1 The importance of driver position. Allen (1962) reported that the average driver tends to sit very near to the left door. This lateral positioning increases the obstruction of the left A-pi1 lar. I t may also increase windshield distortions due to the increased angu 1 ar displacement. Furthermore, in addition to lateral positioning, what drivers can see from a vehicle is greatly influenced by their height (Heath and Finch, 1952). An implication of these studies is that in experiments concerning visibility, driver head position should be measured and/or control 1ed. 2, Mirror placement and design. Mi rrors can affect vi si bi 1 i ty i n at least two ways. First, poor placement of mirrors (inside or outside) can obstruct the driver's line of sight significantly (A1 1 red, 1978). Second, mirrors must be placed to minimize "blind spots" to the rear and sides of the vehicle, and thereby increase the overall field-of-view of the driver (Kaehn, 1976). To a limited degree, the vision problem of windshield damage is similar to other problems that introduce blind spots in the f ield-of-view. 3. Visibility loss due to spectacle frames. Bewley (1969) claims the style and size of spectacle frames worn by drivers may reduce the field of vision below minimum safety qtandards. Furthermore, some frames may interfere both with peripheral and central vision (Smith and Weale, 1966). Neither the absolute or relative levels of this type of interference are known. 4. Windshield wiping systems. During rain the primary factor reducing visibility is the water film on the windshield (Morris, Mounce, Button and Wal ton, 1977). However, wiper mechanisms can a1 so obstruct a driver's vision. They should be designed to avoid obstruction of certain "critical areas," particularly next to the left A-pi1 lar and the central region of the windshield