Soot trapping on GPF, TWC Membrane fabrication on GPF, and soot trapping on TWC membrane have been studied using computer numerical simulation. The results of pressure drop, filtration efficiency, and morphology visualization satisfied the experimental results. The simulation could clarify the trapping phenomena up to 3D-nanoscale resolution, which overcame an experimental limitation. Visualization revealed the particles first trapped on the substrate generating a dendrite structure at the constricted area. Then, it expanded by accumulating flying particles until the bridge structure was completed, which is the same point of 100% filtration efficiency that proved the hypothesis from the experiment.

抄録全体を表示

Near side crashes make up a large component of fatal crashes. Side airbags protect vehicle occupant in near side crashes, but the effect of the side airbag is not well documented in real-world studies. In a previous report, the author analyzed the effect of side airbag and found the most injured location was the torso, and the effect of torso protection airbag was small compared to the seat belt in frontal crashes. But the effect of curtain airbag could not be evaluated due to insufficient data. Additional data has been added to the data set and now the focus is on evaluating head/neck injury and rollover curtain air bags.

抄録全体を表示

Truck platooning is one of the expected systems using automated driving technology based on cooperated adaptive cruise control (CACC) as a solution to problems such as fuel efficiency and lack of driver. To implement the truck platooning system, it is necessary to ensure the safety and acceptability of the system for user and peripheral drivers. In this study, we aim to improve the safety and acceptability of the truck platooning system using a human-machine interface (HMI) to notice the system information. Focusing on the peripheral drivers and traffic density, the hypotheses that communication to peripheral drivers using the HMI improves the safety and acceptability of platooning trucks for peripheral drivers and that traffic density affects them were tested using a driving simulator (DS). We examined driving performance indices and subjective evaluation of peripheral drivers for the scene of merging in highway junction and emergency stop as a minimal risk maneuver. The results showed that early information transmission to peripheral drivers using HMI can improve the safety and acceptability for the peripheral drivers and that high traffic density may weaken the effects of HMI. This suggests that it is necessary to adjust the specification regarding the movement of platooning trucks with consideration for the traffic density on the road.

抄録全体を表示

Auto-estimation of drowsiness of a vehicle driver is an urgent problem, as more than 15% of traffic accidents are caused by careless driving in Japan. This paper considers drowsiness estimation based on the measurements of the heartbeat, respiration, and the center of pressure on the seat. As the learning model for the prediction, we employ an echo state network (ESN). An ESN is one of the recurrent neural networks (RNNs) that solves the problem of the high computational cost of ordinary RNNs by restricting the tunable parameters. As drowsiness at a time is dependent on the previous time, i.e., the system has dynamics, RNNs will be suitable models for drowsiness estimation. Furthermore, because of their low computational cost, ESNs are a promising candidate model for driver drowsiness estimation. The validity of the proposed method is tested by several persons' data.

抄録全体を表示

The number of traffic accidents and the number of fatalities are on a downward trend. On the other hand, the number of traffic accidents involving the elderly is on the increase. One of the reasons for this is the decline in driving ability due to aging. In this study, we focused the characteristics of the decline in driving ability such as the inability to brake sufficiently in an emergency. Specifically, by comparing two sitting postures (user-selected and instructed), we analyzed the characteristics of the accelerator pedal-to-brake pedal movement and the braking force of the elderly. In addition, the characteristics of aging were also analysed by comparing the results with those of younger subjects. The results showed that the step-over time was faster in the instructed posture for both the young and the elderly. The maximum braking force was stronger in the instructed posture for both the young and the elderly. There was no difference in braking force between the young and the elderly.

抄録全体を表示

To increase the acceptance and willingness of customers in using advanced driver assistance systems and autonomous vehicles, one of the best solutions is to reproduce the driving behavior of humans. In this regard, estimating risk based on driver behavior is a priority. An index of perceptual risk estimation by drivers toward pedestrians approaching and overtaking maneuvers is proposed herein. Risk estimation comprises longitudinal and lateral risk estimations. The index is verified using maneuver data obtained from a driving simulator. It is discovered that drivers with different driving styles have different risk index ranges. The index may be used to classify drivers for selecting the appropriate response from an assisting system.

抄録全体を表示

Indoor temperature affects the arousal level and thermal comfort of drivers. In this study, we aim to prevent the drowsy state and to improve the thermal comfort of drivers in driving for a long time using changes in ambient temperature in cabin. Targeting the scene of driving for 90 minutes, the hypothesis that repeated thermal stimulation to the driver can prevent the decrease of arousal level and thermal comfort of the driver was tested. We investigated the drowsiness level, thermal comfort based on subjective evaluation, and physiological parameters of the driver in the condition of repeated thermal stimulation for 90 min with driving in a simulator. As a result, the drowsiness level with a scale from 0 (alert) to 4 (extremely drowsy) was kept below 1 (slightly drowsy), and the comfort sensation with a scale from -3 (very uncomfortable) to 0 (comfortable) was maintained above -1 (slightly uncomfortable). These results suggest that repeated thermal stimulation is valid to prevent the decrease of arousal level and thermal comfort of the driver.

抄録全体を表示

Understanding car driver responses in emergency car-to-cyclist collisions is useful for developing active safety systems to reduce injuries in car-to-cyclist collisions. In this study, driver responses toward cyclists intruding into a car’s path at an intersection were investigated in scenarios with different time-to-collisions (TTC) (1.5 s and 0.8 s for the near side approach, and 2.0 s for the far side approach) using a driving simulator. The drivers’ responses to avoid collisions were braking and swerving, whereas swerving alone was not effective for avoiding collisions. Collisions occurred when the sum of braking reaction time (BRT) and car braking deceleration time was larger than the TTC when the cyclist was visible to drivers. The mean of BRT was 0.65 and 1.35 s in the near-side and far-side scenarios, respectively. The driver’s initial gaze has an important role in affecting the driver’s time to notice the cyclist. The gaze of many drivers was located on the future path (car lane) when the cyclist was visible. As the angle between the gaze direction and the cyclist was larger, the time taken for the driver to notice the cyclist was longer. As this notice time increased, the driver’s response changed consistently from swerving to braking. Through this study, a collision area is proposed using TTC and car velocity based on the collision occurrence equation.

抄録全体を表示

We research parking scene reconstruction by a deep neural network (DNN) using a millimeter-wave radar. High accuracy can be achieved for the training dataset; however, it is degraded on untrained data. It is an unavoidable challenge of generalizability in machine learning. To solve this issue, we propose the method to utilize the model knowledge and estimate the reliability of the reconstruction results. The vehicle shape model with the Bezier curve is applied, and the shape parameters are estimated by DNN. The reliability is estimated by a generative model, Variational Auto Encoder (VAE) trained by the training dataset. When the reliability is low, the estimated model parameter is corrected with a stable deductive model. The correction effectiveness was confirmed for the experimental data measured in actual parking scenes.

抄録全体を表示