Search Results

The purpose of this study was to construct driver models using long short-term memory (LSTM) in car-following situations, where other vehicles change lanes and cut in front of the ego vehicle (EGV). The development of autonomous vehicle systems (AVSs) using personalized driver models based on the individual driving characteristics of drivers is expected to reduce their discomfort with vehicle control systems. The driving characteristics of human drivers must be considered in such AVSs. In this study, we experimentally measured data from the EGV and other vehicles using a driving simulator consisting of a six-axis motion device and turntable. The experimental scenario simulated a traffic congestion scenario on a straight section of a highway, where a cut-in vehicle (CIV) changed lanes from an adjacent lane and entered in between the EGV and preceding vehicle (PRV).

This study aimed to construct driver models for overtaking behavior using long short-term memory (LSTM). During the overtaking maneuver, an ego vehicle changes lanes to the overtaking lane while paying attention to both the preceding vehicle in the travel lane and the following vehicle in the overtaking lane and returns to the travel lane after overtaking the preceding vehicle in the travel lane. This scenario was segregated into four phases in this study: Car-Following, Lane-Change-1, Overtaking, and Lane-Change-2. In the Car-Following phase, the ego vehicle follows the preceding vehicle in the travel lane. Meanwhile, in the Lane-Change-1 phase, the ego vehicle changes from the travel lane to the overtaking lane. Overtaking is the phase in which the ego vehicle in the overtaking lane overtakes the preceding vehicle in the travel lane.

Many automated driving technologies have been developed and are continuing to be implemented for practical use. Among them a driver model is used in automated driving and driver assistance systems to control the longitudinal and lateral directions of the vehicles that reflect the characteristics of individual drivers. To this end, personalized driver models are constructed in this study using long short-term memory (LSTM). The driver models include individual driving characteristics and adapt system control to help minimize discomfort and nuisance to drivers. LSTM is used to construct the driver model, which includes time-series data processing. LSTM models have been used to investigate pedestrian behaviors and develop driver behavior models in previous studies. We measure the driving operation data of the driver using a driving simulator (DS).

Most driving assist systems are uniformly controlled without considering differences in characteristics of individual drivers. Drivers may feel discomfort, nuisance, and stress if the system functions differently from their characteristics. The present study reduced these side effects for systems with a highly accurate driver model. The model was constructed using Nonlinear AutoregRessive with eXogeous inputs (NARX), which has a learning function and estimates the driving action of a driver. The model was constructed for one driving condition yet can be applied to other driving conditions. If one model can be applied to many driving conditions, a system can construct as minimum requirements. The driver decelerated while approaching the target at the tail of a traffic jam on a highway. A driver model was constructed for the driver’s braking action. The experimental condition was 11 data measurements from 50 to 130 km/h made at intervals of 10 km/h.

A vehicle-to-vehicle communication system (V2V) sends and receives vehicle information by wireless communication and assists safe driving. The present study investigated the activation timings of collision information support, collision caution support, and collision warning support provided by a V2V in an experiment using a driving simulator for four situations of (1) assistance in braking, (2) assistance in accelerating, (3) assistance in making a right turn, and (4) assistance in making a left turn at a blind intersection. The four situations are common scenarios of traffic accidents in Japan. Safety margins for collision information support and collision warning support were the time required for the driver to fully apply the brake pedal, while the safety margin for collision caution support was the time required for the driver to begin applying the brake pedal. The study investigated the effects of adding safety margins to standard activation timings.

The purpose of this study is to clarify the mechanism of traffic accidents involving cyclists. The focus is on the characteristics of cyclist accidents and scenarios, because the number of traffic accidents involving cyclists in Tokyo is the highest in Japan. First, dangerous situations in traffic incidents were investigated by collecting data from 304 cyclists in one city in Tokyo using a questionnaire survey. The survey indicated that cyclists used their bicycles generally while commuting to work or school in the morning. Second, the study investigated the characteristics of 250 accident situations involving cyclists that happened in the city using real-world bicycle accident data. The results revealed that the traffic accidents occurred at intersections of local streets, where cyclists collided most often with vehicles during commute time in the morning. Third, cyclists’ behavior was observed at a local street intersection in the morning in the city using video pictures.

Vehicle to vehicle communication system (V2V) can send and receive the vehicle information by wireless communication, and can use as a safety driving assist for driver. Currently, it is investigated to clarify an appropriate activation timing for collision information, caution and warning in Japan. This study focused on the activation timing of collision information (Provide objective information for safe driving to the driver) on V2V, and investigated an effective activation timing of collision information, and the relationship between the activation timing and the accuracy of the vehicle position. This experiment used Driving Simulator. The experimental scenario is four situations of (1) “Assistance for braking”, (2) “Assistance for accelerating”, (3) “Assistance for right turn” and (4) “Assistance for left turn” in blind intersection. The activation timing of collision information based on TTI (Time To Intersection) and TTC (Time To Collision).

Although traffic accidents in Japan involving bicycles have been decreasing yearly, more than 120,000 per year still occur. Few data exist regarding the mechanisms underlying bicycle accidents occurring at intersections. Such dangerous situations form the backdrop of the warning and automatic braking systems being developed for motor vehicles. By clarifying cyclist behavioral characteristics at crucial times, it may be possible to introduce a similar warning system for cyclists as a countermeasure to reduce accidents. The objective of this study is to clarify the mechanism of accidents involving bicycles and to obtain useful data for the development of a warning system for cyclists. A video camera and software investigated and analyzed cyclists' speed and trajectory at an intersection where many accidents occur. Cyclists entering the intersection from one direction were recorded.

Collision avoidance systems for rear-end collisions have been researched and developed. It is necessary to activate collision warnings and automatic braking systems with appropriate timing determined by a monitoring system of a driver's braking action. Although there are various systems to monitor driving behavior, this study aims to create a monitoring system using a driver model. This study was intended to construct a model of a driver's braking action with the Time Delay Neural Network (TDNN). An experimental scenario focuses on rear-end collisions on a highway, such as the driver of a host vehicle controlling the brake to avoid a collision into a leading vehicle in a stationary condition caused by a traffic jam. In order to examine the accuracy of the TDNN model, this study used four parameters: the number of learning, the number of neurons in the hidden layer, the sampling time with 0.01 second as a minimum value, and the number of the delay time.