Capítulo 3. Simbólicas Preliminares
3.5. La Cabellera y la adivinación
A pilot study was designed and implemented to evaluate the effectiveness of the developed ROR training video, and whether participants could improve their ROR recovery skills in a simulator based learning environment. While many factors influence ROR events, this study focused on the general straight road ROR case to simply address these two criteria. The subsequent pilot study design and evaluation methods are presented in the following sections.
5.3.1
Design
In the United States, most ROR scenarios begin with the vehicle exiting the roadway to the right. As a result, the simulation environment for the pilot study was designed to focus on ROR departures on the right side of the road only. The roadway was created in CarSim as a two lane oval track with long straight sections as shown in Figure 5.3. Cones were placed along the centerline of the track to keep drivers in the correct lane and additional cones directed them off the road at specific points as shown in Figure 5.4. Unequal friction coefficients were added to the roadway surface and shoulder to replicate the dangers of a split-mu surface. The coefficient for the roadway surface was set to µr = 0.9, typical for dry pavement, and the shoulder was
set to µs = 0.25, a value close to that found on grass and loose gravel. Lastly, the
roadway surface was raised slightly above the shoulder creating a lip-height of 0.1 meters.
Figure 5.3: ROR test track created in CarSim for the study
The vehicle used in the simulation was a Honda Odyssey minivan with front- wheel drive, standard anti-lock brake system, and independent front and rear sus- pension. The geometry of the minivan along with mass and inertial properties are
Figure 5.4: Road geometry for the straight sections of the CarSim track. The red triangles denote cone placement. Note: Image is not portrayed to scale.
displayed in Table 5.1.
The pilot study structure, as shown in Figure 5.5, began with an introductory video including a brief definition of the term “run-off-road” and some visual examples (Video 1) to familiarize the participant with the theme of the study. Next, the subject received a pre-test questionnaire which collected demographic information and evaluated knowledge and experience with ROR prior to participation in the study. Once the questionnaire was completed, the subject’s present skills in recovering from an ROR event were evaluated. The subject was instructed to follow the cones off the track at about 40 mph (65 km/h) into an ROR situation and was given a single opportunity to safely recover the vehicle back into the correct lane of the roadway.
Following the preliminary evaluation, the training program (Video 2), de- scribed in Section 5.2.2, was viewed by the participant on the simulator screens. When the video finished, the subject was permitted to practice the instructed ma- neuvers on the evaluation track. No limitations were placed on the practice time and the participants simply notified the coordinator when they felt comfortable with the recovery process. A third video was then presented to explain the final evaluation
Table 5.1: Physical properties of the Honda Odyssey used in the CarSim simulation
Symbol Description Value Units
h Height Sprung Mass 1.500 m
Ixx Roll Inertia 729.7 kg m2 Iyy Pitch Inertia 3916.5 kg m2 Izz Yaw Inertia 4029.5 kg m2 Ixy Product of Inertia 0 kg m2 Ixz Product of Inertia 181.5 kg m2 Iyz Product of Inertia 0 kg m2
`f Mass Center to Front Wheel 1.3559 m
`r Mass Center to Rear Wheel 1.6464 m
`w Wheel Base 1.6916 m
m Sprung Mass 1788.8 kg
wt Tire Width 0.205 m
tests. For Test 2, the subjects were instructed to drive off the road at the cones at 60 mph (97 km/h) and recover the vehicle in a safe and controlled manner. The same procedure was then executed again only this time at 40 mph (65 km/h) for Test 3. Lastly, each participant filled out a post-test questionnaire to evaluate their understanding of ROR following the training program.
5.3.2
Evaluation
The results obtained from the pilot study included the subjects’ answers to the two questionnaires and vehicle data obtained from the simulator software during the three tests. CarSim provided a variety of vehicle states and parameters including positions, velocities, accelerations, angular velocities, slip angle, and tire forces. The steering wheel angle was also obtained directly from the dSPACE software. A scor- ing system was developed to systematically combine the results from the study into scoring metrics and provide overall scores to each participant. The scores for each test, Si, were calculated as
Si = 10
X
j=1
aijK¯ij (5.9)
where i is the test number (i = 1, 2, 3), j is the evaluation factor (j = 1, 2, ..., 10), aij
represents the weighting parameter, and ¯Kij is the respective evaluation factor score.
The various assessment factors are located in Table C.3.
The ten evaluation factors were chosen based on the particular results obtained from the study which offered insight into the driver’s understanding and execution of the ROR recovery process. The first two evaluation factors, Q1 and Q2, were the
responses from two questions in the provided questionnaires. The first question (j = 1) was a multiple choice question inquiring how the vehicle should be controlled once an ROR situation has occurred. From the available answers including “applying the brakes”, “maintaining velocity while steering back onto the road”, and “applying the throttle”, the correct answer (B) was to “steer straight and slow down by coasting”. This question evaluated whether the participant understood the safest way to initially react to ROR. The second question (j = 2) asked what factors should be considered prior to recovery from ROR. This question was also multiple choice but the participant was instructed to mark all the correct answers of which there were three. The correct
answers (A,B,D) were “traffic speed limits”, “barriers or oncoming traffic”, and “hand placement on the steering wheel” respectively, while the incorrect answer (C) was “vehicle height”. This question checked the participants’ knowledge on some of the factors which can make ROR recovery very dangerous if not accounted for properly.
Following the questionnaire responses, the participant’s vehicle velocity while attempting to recover, uR, was evaluated (j = 3). During the video training program,
it was recommended that a safe speed to recover the vehicle is below 25 mph (40 km/h). The subjects were awarded the full amount of points for speeds in this range and were incrementally penalized the further their velocity deviated from this range. The next two evaluation factors were the maximum left (θL) and maximum
right (θR) steering wheel angle used during recovery (j = 4 and j = 5, respectively).
The motivation for these factors is that the steering is the primary means of control- ling the vehicle during recovery and steering angles used can greatly influence how safe or dangerous the situation becomes. The video training program explained to the participants that a safe steering maneuver during recovery is to turn the wheel 90 de- grees to the left and then 90 degrees to the right to counter steer. Those subjects who implemented the suggested steering angles within plus or minus 25 degrees received a full score as shown in Table C.3. Less points were awarded for smaller steering angles as this creates a shallow angle of approach and a greater chance of scrubbing the tires on the roadway edge. Steering wheel angles above the recommended range were also penalized because of the dangers of overcorrection.
The slip angle, β, and yaw rate, ˙ψ, were two vehicle states computed by Carsim which are often used to characterize the stability of a vehicle. These two parameters are useful because they influence the amount of traction available between the tires and road. Slip angles and yaw rates above specific thresholds characterize a loss of traction and thus a loss of control of the vehicle. Therefore, the next two evaluation
factors were established as Rβ, the ratio of the slip angle to the safe threshold (j = 6),
and Rψ˙, the ratio of the yaw rate to the safe threshold (j = 7). The safe thresholds for
these two parameters were defined according to [104] as βT hreshold = tan−1(0.02µg)
and ˙ψT hreshold = µgu where µ is the tire-road friction coefficient, g is gravity, and u is
the longitudinal velocity of the vehicle. Full scoring was awarded for ratios below 85 percent with fewer points awarded for ratios closer to and exceeding 100 percent.
One very important evaluation factor, for this study, was the lateral displace- ment (j = 8). The maximum lateral displacement, ymax, provides a clear indication
of whether the participant was able to recover the vehicle back into the proper lane without crossing any of the cones in the simulation. The full number of points were awarded if -3.15 m < ymax < -0.85 m as these values corresponded to the vehicle’s
placement in the correct lane (see Figure 5.4). Positions greater than -0.85 m indi- cated that the vehicle traveled into the oncoming lane creating what would be a very dangerous situation and were awarded no points. Maximum lateral displacements below -3.15 m indicated that the vehicle was never recovered back onto the roadway or got stuck in a two-wheels-off position. This condition is also dangerous and was not awarded any points given the high potential for collision with obstacles and unstable surfaces on the side of the road.
The ninth evaluation factor (j = 9) was a simple assessment of whether the participant scrubbed the tires of the vehicle against the roadway edge during recovery. Scrubbing the tires is a common and dangerous mistake during recovery, often leading to the “slingshot” effect discussed previously. Scrubbing can also completely prevent the vehicle from being able to mount the surface and return to the road so no points were awarded if this condition occurred.
The last evaluation factor (j = 10) was whether or not the participant was able to recover the vehicle back into the proper lane by the end of the roadway space
provided. A full score was awarded if the vehicle data showed controlled motion in the right lane no matter how safe the recovery maneuver.
To separately evaluate the effectiveness of the training video and whether the subjects’ skills improved, the results were distinguished according to two categories. The first category, “Training” factors, included results which directly demonstrated the effectiveness of the training program. During training, the subjects were pro- vided with specific instructions on how to control and recover a vehicle back onto the roadway including proper steering angles and speed. Thus results which matched the “Training” category were the responses from the questionnaires, the longitudinal velocity and the maximum left and right steering wheel angle. The second cate- gory, “Safety” factors, included vehicle parameters and simulation occurrences which characterized the safeness of the recovery maneuver. The results which matched this category were the slip angle and yaw rate ratios, the lateral displacement, tire scrubbing, and whether the vehicle recovered.
Finally, the overall improvement of each participant was of interest in analyzing the effects and success of the ROR training program. To quantify improvement, each participant received a performance score based on the following formula.
SIM P =
1
2(S2+ S3) − S1 (5.10)