LED NAVIGATION SYSTEM

Transcription

1 Zachary Cook Adam Downey LED NAVIGATION SYSTEM Aaron Lecomte Meredith Swanson UNIVERSITY OF NEW HAMPSHIRE DURHAM, NH Tina Tomazewski Andrew Kun ABSTRACT Research shows that when drivers use a personal navigation device (PND), they spend less time focusing on the road. To prevent the driver from having to take their eyes off the road, a peripheral navigation system could be implemented above the driver s head. This would allow the driver to view driving directions while they focus on the road ahead. The peripheral navigation system will consist of an LED array connected to a microcontroller and a button press keypad to display the directions to the driver. The new system was tested in conjunction with a standard personal navigation device (PND) to provide baseline data for comparison. After analyzing the data, the standard PND had a percent dwell time (PDT) of 85% while the LED Navigation System had a PDT of 94%; a 9% increase. This may seem insignificant but after an hour of driving that 9% equates to approximately 6 extra minutes that the driver was focused on the road. The evaluation of the exit survey led to the conclusion that 7 out of 10 participants preferred the LED navigation system over the standard PND. Author Keywords Driving simulation, LED Navigation, Standard PND ACM Classification Keywords H5.m. Information interfaces and presentation General Terms Measurement, Design, Performance, Experimentation, Human Factors. INTRODUCTION In some countries driving is the primary mode of commuting. For example, according to the U.S. Census Bureau [4], Americans spend more than 100 hours a year commuting to work. Given the large amount of time that people spend behind the wheel, and the increasing availability of computational resources that can now operate inside a vehicle, companies have been introducing myriad mobile services and functionalities into the consumer market just for drivers. A few notable examples are hands-free voice dialing, live traffic reports, automated directory assistance, infotainment systems, and personal navigation devices (PNDs). Unfortunately, the question of how these in-car services impact driving performance is often left unanswered. The focus of this paper is the impact of PNDs on visual attention and driving performance. In recent work, Kun explored the effects of two PNDs on visual attention and driving performance [3]. One was a standard PND that displays a map and provides turn-by-turn instructions by voice. As technology advances, drivers find themselves frequently taking their eyes off the road. Using personal navigation devices (PNDs) cause s drivers to be more concerned with following directions rather than focusing on the road ahead. As stated in Glancing at Personal Navigation Devices Can Affect Driving [3], the percent dwell time (PDT) spent paying attention to the road was 90.4% when using a standard personal navigation device. By designing a peripheral directional system, we anticipate to increase the driver s PDT spent looking at the road. Through the implementation of a series of LED arrays, navigation instructions will be displayed in the driver s line of sight. This will permit the driver to stay continuously focused on the road while maintaining the desired directional route. RELATED RESEARCH Practically all commercially available PNDs rely on a map displayed on a HDD with a navigation route overlaid on top of it. The presence of the screen entices drivers to look away from the road ahead by Kun et al. [3]. As demonstrated in a driving simulator experiment by Horrey et al. (2006) [2], who explored the effect of interacting with in-car devices on driving performance and visual attention, as users spent less time looking at the road ahead, their lane position variability increased. EXPERIMENT We now describe our experimental setup and procedure. We also detail each different type of PND that was used in our study as well as how we computed independent variables. Equipment The experiment was performed in a high-fidelity driving simulator (Figure 1) with a 180 field of view screen and a full-width automobile cab. The cab sits on top of a motion base which simulates car movements for braking

2 Figure 1: LED Navigation Array and accelerating, as well as bumps on the road. As shown in Figure 2, the simulator was equipped with two sets of eye-trackers which track subjects gaze and head position using two pairs of cameras mounted on the dashboard. Figure 2 also shows the location of the in-car LCD screen which was used as the display for the standard PND. The LCD screen was mounted in the cars console, which is a common place for contemporary PNDs and smart phones with navigation capabilities. The size of the LCD screen was 8 inches diagonally which falls within the typical size of commercially available navigation devices. The LED Navigation System, with a size of 11.5 inches diagonally, is located just above the dashboard. This location is optimal for allowing the driver to focus on the road while receiving directional information via their peripheral vision. The LED navigation system consists of an array of 56 LEDs. These LEDs are illuminated to assist the driver in navigating a driving route during the experiment. The LED Navigation System is controlled by an Arduino microcontroller. All of the potential directions and scenarios were loaded onto the Arduino prior to running experiments. Navigation commands, which consisted of left and right turns, were implemented by pushing their respective button on the keypad, which can be seen in figure 4. The keypad is connected to the Arduino microcontroller which operates the LED Array. Figure 2: Right Turn Sequence 1 Figure 3: Right Turn Sequence 2 Figure 4: Keypad Implementation Left and right turns were implemented by illuminating either the left or right half of the LED Array. The illuminated half, seen in figure 2 displays the beginning of a right turn. This portion of the array flashed intermittently and its frequency increased as the driver approached the upcoming turn. At the turning point, the right portion of the array began to sweep to the right, seen in figure 3. The sweeping motion starts from the middle of the array and moves outward, column by column, either to the right or left depending on the turn needed. This motion repeats itself for 5 seconds. Keypad Function The keypad sends a command to the microcontroller to display the desired turn throughout the simulation. One button projects a right turn and the other projects a left turn. The buttons were pressed at the experimenter s discretion based on the driving style of the participant. As the driver approached a turn, specific landmarks were used as guidelines to notify the experimenter of when to press the button. Before running experiments we compiled a list of all the turning points throughout both routes. With this list we practiced the timing of

3 each route to prevent any mistakes while running an official experiment. Fabrication of Hardware List of materials: 8 x12 X0.5 sheet of lexan x1 bars of acrylic mcd LEDs P(preferably clear) 56 LED plastic sockets (support) Blue painter s tape 6 machine screws The system had to be transparent so that it would not inhibit the driver s field of view. Increasing the PDT on the road was the most important parameter during the design process. Lexan was used for construction due to its resistance to chipping, scratching, and yellowing over time. Using lexan is not the only viable material for the display, but other materials have a greater tendency to crack or split from industrial use. Blue painter s tape was used to cover the lexan sheet to further protect the display from the construction process. The design specifications for the LED display needed to be roughly 10 X6. A hydraulic press was used to cleanly cut the lexan to meet those size specifications. After the lexan sheet was cut, a micrometer was used to accurately space the LEDs evenly across the center of the display in a matrix format. A drill press was used to create a hole for each LED. Each hole had to be slightly smaller than the LED so that the LED would fit snugly. A 3 acrylic bar was used to create the frame for the LED array. The rounded support bars were then mounted to the frame. With the support legs in place, the viewing angle of the display could be adjusted to be appropriately mounted in the simulator. The LEDs were controlled by a matrix which was created by connecting both the columns and rows to the Arduino microcontroller. The columns are connected to the LEDs anodes which results in the columns needing to be high for any of the LEDs in that column to turn on. The rows are connected to the LEDs cathodes which results in the rows needing to be low for an individual LED to turn on. If the row and the column are either both high or both low, no voltage will flow through the LED causing it to remain off. The patterns used in this experiment were created by sending different signals from the microcontroller to the LED array. Figure 5: Fabrication of Display Figure 6: LED Matrix Schematic Method Participants Ten university students participated in the experiment. They were between 18 and 24 years of age (mean age 20 years, standard deviation 1.2 years). As compensation, each received a $20 gift card to a popular store chain. Navigation Aids Each participant performed navigation with each of the following devices: 1. LED Navigation System: Navigation directions were presented by showing turn signals on an LED array located above the instrument panel in the driver s line of site. The location of the LED array can be seen in figure 2. The LED array is capable of showing a straight path, left, and right turns. The LED array alerted drivers Figure 7: LED Navigation System

4 Eye Tracking Cameras Figure 9: Standard Navigational Device Figure 8: Experimental setup inside the vehicle that a turn was approaching by flashing the LEDs in the sequence that showed either a left or right turn. 3. Standard Navigational Device: The participants received navigation directions via a standard consumer style GPS. This navigation device was placed on the dashboard, to the right of the instrumentation panel. This location was chosen to simulate the scenario most commonly used by drivers. Procedure After filling out the consent forms and personal information questionnaires, participants were given an overview of the driving simulator and descriptions of the two navigation conditions. Next they proceeded to complete two navigation experiments, one with each of the PNDs. Before each condition, we provided subjects with about 5 minutes of training using that particular PND. For training, users followed PND navigation instructions in a city environment. In order to circumvent order effects, we counterbalanced the presentation order of the PNDs between subjects. Figure 10: Virtual Route Navigated by the Subjects As shown in Figure 4, subjects drove on two lane city roads which included ambient vehicles, moving pedestrians, traffic signs and lane markings. Lanes were 3.6 meters wide. Subjects were instructed to drive as they normally would in real life and to obey all traffic laws. For each of the PNDs, the participants navigated along a different route. Figure 4 shows the first route used in this experiment. For the second route we reversed the direction of travel. In short, the routes were of the same length (about 10 km) and complexity. However, the turn-by-turn directions for each route were different. Thus, there was no risk of subjects remembering navigation instructions from the previous route. Each route had both long (400 and 800 meter long) and short (200 meter long) segments with many intersections on the given path. On average it took about 15 minutes to

5 traverse a route. The presentation order of routes was the same for all subjects. Finally, at the end of the experiment, subjects ranked their level of agreement with various statements pertaining to the PNDs and provided written feedback about the experiment. Design We chose a within-subjects experiment that was comprised of testing the LED Navigation System and consumer style GPS as independent variables. We measured multiple dependent variables: Percent dwell time (PDT) on the road ahead, which measures the percent of time drivers spend looking at the road ahead. A low value indicates that a driver is distracted, which in turn can lead to collisions. Average driving performance measures, which included variances of lane position, steering wheel angle and velocity. In each case, higher values for driving performance measures indicate deterioration in performance. Level of agreement with preferential statements, which reflects subjective opinion about PNDs, was collected using a 5-point Likert scale. Measurement Using an eye-tracker we were able to automatically classify gazes as being directed at the road ahead, at the navigation device, or somewhere else inside the cabin (e.g. the speedometer). The sampling frequency of the eye-tracker was 60 Hz. In rare occasions when the eyetracker was not able to resolve where the subject was directing his/her gaze, the experimenter made classifications using video footage from a camera installed on the dashboard. Lane position and steering wheel angle were recorded by our driving simulator software at a frequency of 10 Hz. Calculation Data segmentation: The city routes in our experiment can be broken up into segments by treating roads between two intersections as separate segments. The segments can be seen in figure 4.We calculated the PDT, lane position variance, and steering wheel angle using the values logged by the simulator. All segments had the same characteristics, thereby controlling factors that could potentially confound our results. Participants did not encounter any unexpected events, which often require sudden braking and steering wheel motion. This in turn can result in very large first differences and variances for these measures, making comparisons with other segments difficult. In analyzing all of the segments, we excluded data collected over the first 60 meters and the final 40 meters of a segment, and analyzed data generated over ( ) = 100 meters. This was done because driving performance is different between the excluded and analyzed portions of the segments. For example, at the beginning of a segment, drivers are completing the turning maneuver that is necessary to get through the previous intersection. And at the end of a segment, they are decelerating before entering the next intersection. Thus, the resulting first differences and variances can be much larger than those encountered away from intersections, which makes it difficult to compare excluded and analyzed portions of segments. Using the aforementioned short segments, we calculated the following measures. Visual attention: For each participant p and navigation aid nav, we calculated the average percent dwell time, PDT p,nav on the road ahead by finding the ratio of the sum of dwell times for all 13 segments and the sum of the total time spent traversing all 13 segments. We defined looking at the road ahead as looking at one of the three projection screens of the simulator. Similarly, we calculated the percent dwell time on the LCD screen when participants used the SV and SPND. Finally, we calculated the percent dwell time on the rest of the cabin (speedometer, rear view mirror, etc.). Average driving performance measures: For each participant, segment, and navigation type, we calculated the average absolute values of first differences and the variances of the following three driving performance measures: lane position, steering wheel angle and velocity. We also calculated the average velocity for each segment. For each participant and navigation aid we then calculated the averages for these variables over the route. Subjective assessment: Agreement levels with preferential statements and written comments were transcribed from paper forms. We also solicited qualitative verbal comments about the experiment from participants. RESULTS Visual Attention Table 1 shows the PDT on the road ahead. We conducted a repeated measures ANOVA to assess the effect of different PNDs on visual attention using PDT

6 on the road ahead as the independent variable. The analysis revealed a significant main effect on PDT (p < 0.05). PDT for the LED display was the highest at 95%, while the standard GPS had a PDT of 85%. Post-hoc comparisons indicated a significant difference between the LED display and standard GPS (p = 0.001). As hypothesized, PDT increased when using the LED display. Our results seem to indicate that keeping gaze between the standard GPS and the observed world takes more time and effort. LED Dispay Standard GPS p-value <0.05 Table 1: PDT for LED Display vs. Standard GPS Driving Performance With an increase in PDT from using our LED display, we correspondingly observed an overall improvement in driving performance. The driving performance parameters analyzed to support this were lane position variance and steering wheel angle variance. Table 2 shows these values for both the GPS and LED display. For both parameters, less variance occurred with the LED display which is indicative of better driving performance. After performing statistical analysis, we found significant p values of p=0.006 for lane position variance and p=0.035 for steering wheel angle variance. Parameter Analyzed PND used Mean Value p-value Lane Position Variance (m^2) Steering Wheel Angle Variance (degrees^2) LED Display Standard GPS LED Display 7.61 Standard GPS Table 2: Driving Performance for LED Display vs. Standard GPS According to our results, glancing away from the road ahead does affect driving performance. Subjective Assessment Preferential Statements Table 3 shows the percentage of participants who agreed or disagreed with two preferential statements. Responses to these statements indicate that subjects liked using the LED display, and that most of them felt they used only their peripheral vision to follow directions. Statement Agreement Response Percentage I preferred the LED highly agree or agree 70% display over the standard highly disagree or GPS 30% disagree I used my peripheral vision instead of looking at the LED display highly agree or agree 60% highly disagree or disagree Table 3: Responses to Preferential Statements Table 4 shows a comparison of another preferential statement to determine how distracted subjects felt using both navigation types. Note that for this question, the percentage does not add up to 100% since some subjects had a neutral opinion. Based off of our p value that was found using the Wilcoxon Signed Rank test, there was not a significant difference in which system was found more distracting. Table 4: Comparison of Distraction Felt for each System 30% Statement Agreement LED Display % Standard GPS % p-value I found the system to be distacting Written Comments Seven out of ten participants provided either written or verbal feedback that indicated a preference for the LED display over the standard GPS. Quote 1 (Q1) of a written comment by participant 1 shows that they found the LED display easy to use and was able to obtain directions without glancing up at it. Q1 P(1): highly agree or agree highly disagree or disagree 10% 50% 70% 20% "The LEDs were a lot easier to follow since I didn't have to go out of my way to look at it." Another participant, P(7), agreed with the previous participant that the LED display was easier to use, and further added that they would personally consider purchasing something like this if it were closely priced to a commercially available GPS. Q2 P(7): "I saw improvement in reaction times when using an LED display. It was significantly easier to use and I would consider using it myself if it was similarly priced." Although most participants liked and preferred the LED display, some pointed out aspects they disliked about it and made further suggestions. Quote 3 is from

7 participant (5), and shows that they like the LED display but would still like to have the feel of a standard GPS. Participant (12) provided us with our fourth quote, pointing out several features of the display they found issues with. This participant was excluded from the study for data processing due to poor driving skills. However, this participant had the most negative comment, which points out that there is more work to be done to improve the system. Q3 P(5): Q4 P(12): "I suggest adding the LED in addition to a standard GPS display." "Too high up, too bright, no indication of future turns. Flashing distacting." DISCUSSION We started this study by proposing a hypothesis. We will now consider that hypothesis in light of our results 1. An LED navigation system placed above the driver s line of sight will increase Percent Dwell Time (PDT) and improve driving performance Participants average percent dwell time (PDT) on the road ahead was higher for the LED display (94%) than the standard GPS (85%). Clearly, the LED display allowed users to focus on the road ahead instead of being distracted by the standard GPS. Participants variance of lane position as well as steering wheel angle were both lower when using the LED display instead of the standard GPS. Lane position variance averaged m 2 for the LED display and m 2 for the standard GPS. Steering wheel angle variance averaged 7.61 degrees 2 for the LED display and degrees 2 for the standard GPS. The fact that we observed an increase in PDT and an improvement in driving performance suggests that using a peripheral means of navigation, an LED display in our case, may be a better alternative to the standard GPS. Subjective assessments also favored the LED display. When reviewing preferential statements, participants indicated that they preferred the LED display over the standard GPS and felt that they used only their peripheral vision. demonstrated how this increase in visual attention was associated with better driving performance over the standard GPS. More importantly, participants subjective assessments of the LED display were largely positive. For future research, possible changes to the LED display could be explored to see the effects on visual attention and driving performance. This includes the change in location of the LED display, adjusting the illumination of the LEDs, and the change in directional pattern of the LEDs. The LED display could also act as an addition to the standard GPS, which would allow users to obtain both peripheral directions from the LEDs and the physical route on the standard GPS. REFERENCES 1. Burnett, G. E. A Road-Based Evaluation of a Head- Up Display for Presenting Navigation Information. In Proc. HCI Int l Conference (2003). 2. Horrey, W. J., Wickens, C. D., and Consalus, K. P. Modeling Drivers' Visual Attention Allocation while Interacting with In-Vehicle Technologies. Experimental Psychology: Applied 12, (2006), Kun, A. L., Paek, T., Medenica, Ž., Memarović, N., and Palinko, O. Glancing at personal navigation devices can affect driving: experimental results and design implications. In Proc. AutomotiveUI '09, ACM Press (2009), Lowe, S. Many Workers Have Long Commutes to Work. US Census Bureau Press Release. /archives/american_community_survey_acs/ html. March 29, Retrieved 5/30/ Medenica, Ž., Kun, A., Paek, T., Palinko, O., Augmented Realiy vs. Street Views: A Driving Simulator Study Comparing Two Emerging Navigation Aids. (2011) Conclusion & Future Direction This paper presents a thorough driving simulator based comparison of two types of PNDs. Using a simulated city environment in a high fidelity driving simulator, we found that the LED display yielded more visual attention on the road ahead as compared to the standard GPS. When using the LED display, participants spent 5.4 sec more each minute looking at the road ahead as compared to the standard GPS. Furthermore, we