In this research, we have proposed a passive front wheel steering that mechanically follows the vehicle body tilting for narrow track vehicle. The main purpose of this work is to investigate the steering stability of the vehicle with passive front wheel design. A three-wheel vehicle model was developed based on multibody system approach for dynamics simulation and analysis. We have conducted three steering stability analyses by simulation: （a） Low speed steering analysis; （b）Slalom steering analysis; （c） Uneven road surface steering analysis. A prototype vehicle was developed to conduct experiment for comparison to validate the simulation modeling development. Three comparison analyses are illustrated: （i）Constant circular turning analysis; （ii） Single lane change analysis; （iii） Rolling frequency and phase shift analysis. The simulation results are closely agreed with the experiment results, proven our vehicle model can simulate accurately the vehicle dynamics with passive front wheel steering.
Sensors play an important role in recognizing the traffic situation around the vehicle. Pedestrian detection systems using optical sensors or millimeter-wave radars have become widespread, but it is not easy to detect pedestrians in the shadows such as blind corners. In this research, we propose a pedestrian detection method using space potential fluctuation around the human body caused by contact separation between the shoe sole and the road surface during walking. And by simulation and experiment investigated the possibility of pedestrian detection in the blind corner.
IIt is important for traffic management to understand unusual conditions such as road abnormality caused by disasters such as earthquake and heavy rain, and traffic congestions caused by sightseeing and large events. For sensing such situations, various kinds of systems have been developed and used in practice. Moreover, further understanding can be expected if a driving video recorder equipped on a car is exploited as a sensor, which rapidly became wide use in daily life. We focused on the flooded road as one of the unusual events and proposed a method to detect it automatically, using deep learning method from driving videos. Since such unusual events hardly occur, amount of training data for the learning is not sufficient. We aim to solve this problem by transfer learning using images taken not only from on-vehicle camera but also from roadside, and synthesized images by CG. We created dataset and verified the performance of detecting flooded scenes.
MaaS （Mobility-as-a-Service） originated in Finland is now making great impacts on transport societies around the world. Even in Japan without transportation unions for passenger-oriented fare adjustment scheme, movements to enable mutual use among different transportation-IC cards and to distribute basic data among public transport modes have been seen. This paper summarizes activity profiles of the Kashiwa ITS promotion council for some 10 years, and future prospects in Kashiwa city towards “Japanese Maas”. This is abridged translation of abstracts for oral presentation at WCTR Society SIG G2 Vienna in 2018.
This paper clarifies the relationship between a Macroscopic Fundamental Diagram (MFD) and congestion patterns on a unidirectional network. Specifically, we formulate a new inverse problem of the dynamic user equilibrium assignment problem for a given congestion pattern. The proposed problem is formulated as a system of linear equations; by solving this, we can derive an analytical formula of a trip completion rate consistent with the congestion pattern.
Identification of the domestic and international impact on society and industry and the risks that will result from changes, as well as measures to mitigate impact and risks, and formulation of scenarios from a long-term perspective will be needed in order to develop more advanced automated driving systems and ensure their widespread use. Therefore, its social needs are clarified, technical development scenario is formulated, and social impacts of automated driving system, negative impact, challenges and scenarios to address these challenges are studied.
Recently research and development of automated vehicle has been developed, and one of the driving environments for deployment of automated vehicle is expressway. In accordance with this, it is necessary to study road operation policy under taking into account automated vehicle in order to utilize expressway’s network efficiently. And in case of no precedent policy, it is also necessary to evaluate this policy in advance using virtual traffic experimental environment such as driving simulator. In this study new road operation policy for smooth traffic using automated vehicle was studied, and items which should be evaluated, function and performance required of virtual traffic experimental environment were studied.
In recent years, significant progress of highway express inter-city buses has been made, and considering the future task of further improvement of safety and services are important. Moreover, the luck of bus drivers is social problem. Now, automated driving and platoon operation are under development with industry, government and academia collaborations. It is considered beneficial to introduce the automated system to express bus. In this paper, we consider the concept for implementation, based on the Roadmap by the Japanese Cabinet and the truck platooning technical development status of industry-academia collaboration under the project by METI and MLIT.
For the cognition of the driver after take over request from automatic driving to manual, a field experiment was conducted. The subjects simulated the operation of automatic driving and manual driving by using a game handle as a dummy at the passenger side. We examined the effect of HMIs using head-up display （HUD） or warning sound for peripheral contingencies during driving. During the manual operation right after the authority transition, we assumed the situations of overtaken by another car from the adjacent lane, sudden deceleration of a leading vehicle, and pedestrians likely to cross over the road. We evaluated differences in cognitive processing by focal distance of HUD images and influence of three-dimensional audio on driving behavior.
Recent advancement in measurement techniques such as bioimaging technologies enables mathematical understanding of biological phenomena based on quantitative data. In this review paper, through our research on immunological ligand discrimination, we show how the quantitative experiments and the hypotheses described by mathematical models have promoted the understanding of the immunological phenomena. We also discuss the future direction of quantitative immunology research.
In this paper, we review a computation paradigm known as reservoir computing. We first show that to perform effective computation, the reservoir should work in a parameter regime that exhibits chaos-like dynamics. On the other hand, harnessing such dynamics requires tuning of not only the output weights but also the reservoir per se. We then introduce three often-used techniques from the machine learning perspective for this purpose. Furthermore, we demonstrate the effectiveness of these techniques by using a number of tasks.
Photoplethysmogram (PPG) for several decades has been commonly used in health monitoring area for the heart rates and oxygen saturation estimations. It has been well researched in terms of these physiological parameters. However, only a limited effort was done towards the investigation of the PPG dynamics, which appears to be highly complex. The methods of nonlinear dynamics were previously used to investigate it. As new types of the PPG have been recently proposed its dynamics need to be carefully studied. This paper gives an overview of the methods of nonlinear time series analysis previously applied to the PPG for its dynamics investigation, and the related results achieved so far.
We review nonlinear time series analysis based on embedding, especially time series prediction and its application to flood forecasting. When we apply the nonlinear time series analysis, we assume that observed time series are deterministically generated by a certain dynamical system, not by a probabilistic distribution.Although it seems difficult to estimate the original dynamical system, we can reconstruct an attractor of the original system based on embedding theorems. Using the reconstructed attractors, we can analyze the dynamics, e.g., detecting causality, predicting future states. In this article, we review classical methods to state-of-the art methods, and we also introduce our ongoing work.
Aging social infrastructure in urban area and huge scale of natural disaster have brought about occurrence of many sinkhole accidents in recent years. Quick and accurate road diagnosis for wide area that serves for prediction and prevention of these accidents is comprehensively expected. This paper describes 1） SKELE-KA technology which use microwave reflection imaging diagnosis for detection of underground cavity, 2） application of Deep Learning procedure for identification of typical reflection image of underground cavity, and 3） effectiveness of Deep Learning procedure for this diagnosis.
A spiking neural network (SNN) is a model that is inspired by information processing in the brains. SNN processes information with action potentials, or spikes. Recently, studies on the deep learning for SNN have been investigated because it could provide us a new powerful information processing tool. Because introducing conventional deep learning algorithms to SNN is mathematically difficult, several techniques that enable those introductions have been proposed. In this review, we introduce several deep learning algorithms in SNN for supervised learning and unsupervised learning. As for supervised learning, the error backpropagation algorithms are explained, while for unsupervised learning algorithms based on spiketime-dependent plasticity are explained.
The precursors of a critical transition (CT) in dynamical systems are called early-warning signals (EWS). The EWS are theoretically derived from the bifurcation theory of dynamical systems. This article introduces the concepts of the EWS through its relationship with bifurcation. I explain the theoretical background of covariance-based EWS as an example. Finally, I conclude this article with the future perspective of EWS studies.
The correlation between the compressive strength of hardened cement paste specimens and pore volume for drilling powder, determined via mercury intrusion porosimetry, was investigated. Only the pores with diameters of 10–300 nm were considered for the pore volume to exclude the void between the particles, based on the results of the particle size distribution measurement of drilling powder. It was confirmed that the pore volume pertaining to pores with diameters of 10–300 nm for drilling powder and the compressive strength of the specimen are highly correlated regardless of the water to cement ratio and curing condition, which indicated that the information of dominant pores over the compressive strength of specimen remained even in drilling powder