Development of Road Surface Detection Algorithm Using CycleGAN-Augmented Dataset.

Road surface detection is important for safely driving autonomous vehicles. This is because the knowledge of road surface conditions, in particular, dry, wet, and snowy surfaces, should be considered for driving control of autonomous vehicles. With the rise of deep learning technology, road surface detection methods using deep neural networks (DNN) have been widely used for developing road surface detection algorithms. To apply DNN in road surface detection, the dataset should be large and well-balanced for accurate and robust performance. However, most of the images of road surfaces obtained through usual data collection processes are not well-balanced. Most of the collected surface images tend to be of dry surfaces because road surface conditions are highly correlated with weather conditions. This could be a challenge in developing road surface detection algorithms. This paper proposes a method to balance the imbalanced dataset using CycleGAN to improve the performance of a road surface detection algorithm. CycleGAN was used to artificially generate images of wet and snow-covered roads. The road surface detection algorithm trained using the CycleGAN-augmented dataset had a better IoU than the method using imbalanced basic datasets. This result shows that CycleGAN-generated images can be used as datasets for road surface detection to improve the performance of DNN, and this method can help make the data acquisition process easy.

Position Estimator Design for a MEMS Top-Drive Electrostatic Rotary Actuator.

The capacitance and rotor angle of a MEMS top-drive electrostatic rotary actuator do not have a linear relationship due to the non-ignorable fringe effect and low aspect ratio of the electrodes. Therefore, the position estimation is not as straightforward as that for a comb-drive linear actuator or a side-drive rotary actuator. The reason is that the capacitance is a nonlinear and periodic function of the rotor angle and is affected by the three-phase input voltages. Therefore, it cannot be approximated as a simple two-plate capacitor. Sensing the capacitance between a rotor and a stator is another challenge. The capacitance can be measured in the electrodes (stators), but the electrodes also have to perform actuation, so a method is needed to combine actuation and sensing. In this study, a nonlinear capacitance model was derived as a data-driven model that effectively represents the nonlinear capacitance with sufficient accuracy. To measure the capacitance accurately, the stator parts for actuation and those for sensing are separated. Using the nonlinear model and the capacitance measurement, an unscented Kalman filter was designed to mitigate the large estimation error due to the periodic nonlinearity. The proposed method shows stable and accurate estimation that cannot be achieved with a simple two-plate capacitor model. The proposed approach can be applied to a similar system with highly nonlinear capacitance.

Design of a Human Evaluator Model for the Ride Comfort of Vehicle on a Speed Bump Using a Neural Artistic Style Extraction.

The subjective evaluation of vehicle ride comfort is costly and time-consuming but is crucial for vehicle development. To reduce the cost and time, the objectification of subjective evaluation has been widely studied, and most of the approaches use a regression model between objective metrics and subjective ratings. However, the accuracy of these approaches is highly dependent on the selection of the objective metrics. In most of the methods, it is not clear that the selected metrics are sufficiently significant or whether all significant metrics are included in the selection. This paper presents a method to build a correlation model between measurements and subjective evaluations without using predefined features or objective metrics. A numerical representation of ride comfort was extracted from raw signals based on the idea of the artistic style transfer method. The correlation model was designed based on the extracted numerical representation and subjective ratings. The model has a much better accuracy than any other correlation models in the literature. This better accuracy is contributed to not only by using a neural network, but also by the extraction of the numerical representation of ride comfort using a pre-trained neural network.

IMU-Based Virtual Road Profile Sensor for Vehicle Localization.

A road profile can be a good reference feature for vehicle localization when a Global Positioning System signal is unavailable. However, cost effective and compact devices measuring road profiles are not available for production vehicles. This paper presents a longitudinal road profile estimation method as a virtual sensor for vehicle localization without using bulky and expensive sensor systems. An inertial measurement unit installed in the vehicle provides filtered signals of the vehicle's responses to the longitudinal road profile. A disturbance observer was designed to extract the characteristic features of the road profile from the signals measured by the inertial measurement unit. Design synthesis based on a Kalman filter was used for the observer design. A nonlinear damper is explicitly considered to improve the estimation accuracy. Virtual measurement signals are introduced for observability. The suggested methodology estimates the road profile that is sufficiently accurate for localization. Based on the estimated longitudinal road profile, we generated spectrogram plots as the features for localization. The localization is realized by matching the spectrogram plot with pre-indexed plots. The localization using the estimated road profile shows a few meters accuracy, suggesting a possible road profile estimation method as an alternative sensor for vehicle localization.