This research proposes a real-time road surface profile estimation method ahead of the vehicle using a stereo camera for preview suspension control. The Performance of suspension control system can be enhanced significantly with the information of road surface profile preview. We verified the estimation accuracy of road surface profile using the proposed image processing method and the test drive shows that the ride comfort can be improved by preview semi-active suspension control system.
From the viewpoint of energy-efficient vehicles and emission problems, lightweight electric vehicles are expected to be spread. However, lightweight electric vehicles are sensitive to loading condition and tend to make drivers control the vehicle in more difficult manner. As a lightweight electric vehicle equipped with in-wheel motors which can independently control each drive wheel torque, Direct Yaw-moment Control (DYC) is effective as a method for improving handling and stability. This research proposes DYC to compensate the changes in the vehicle handling dynamics due to variation in loading conditions. The DYC system is composed of a feedforward of the steering angular velocity and a feedback of the yaw rate. Experiments on the double-lane-change test with a driving simulator were conducted to verify the effectiveness of the proposed DYC system.
As the demand for road freight transportation increases, articulated vehicles (semi-trailer) have the advantage that the maximum load capacity is large. However, there are some problems that high driving skill is required because it is necessary to perform unique steering during backward parking. This research proposes the concept of the parking assist system and path tracking controller. The control system consists of a Pure Pursuit Motion Planning method to determine the reference path tracking and a feedback controller for stabilization of the articulation angle. The experiment using an actual semi-trailer vehicle was carried out and the effectiveness of the proposed approach was verified by evaluating the precision of the parking position.
Shared control between a human driver and a machine is an important issue of human-centered automated driving technology. This research proposes a new shared control system which combines haptic steering control torque together with velocity control in order to conduct stable and appropriate operating guidance resulting in the same way of an expert driver. This system has two following features. The first feature is a reference driver model which calculates the reference steering wheel angle and acceleration/deceleration command simultaneously by using sequential reference path curvature generation. The second feature is the haptic shared control torque controller in which the intensity of the haptic control torque gain are adjusted by using the key concept of the shared ratio (SR). From the driving simulator experiment employed ten subject drivers, the effectiveness of the proposed system in the curved path tracking driving scenario is verified from the viewpoint of path tracking performance and the steering angular velocity.
Risk prediction relevant to pedestrians darting out to roadways is a promising key technology to reduce the number of crashes with pedestrians in unsignalized intersections. The goal of this research is to develop an advanced driver assistance system for preventing crashes with pedestrians in unsignalized intersections by considering knowledge in risk prediction. The reference longitudinal driver model with risk predictive driving behavior is constructed based on the driving data of driving instructors recorded by the continuous logging drive recorder. The driver speed control model is embedded in the proposed driver assistance system and the effectiveness of the driver assistance system is verified.
This research aims to develop an “Autonomous Driving Intelligence System” to prevent risk of accidents and enhance driving safety while keeping human drivers in the control loop. The proposed system focuses on two key technologies: Risk-predictive driving intelligence model and Shared control between the driver and the assistance system. The first key technology is to embed an experienced driver model for recovering degraded performances of recognition, decision-making and operation of drivers. In the driver assistance system design, the experienced driver model contains knowledge-based “risk-prediction mechanism” to avoid accidents in risky driving situations which the system is activated in early stage to assist safe driving when the collision risk becomes imminent. The second key point is “Shared control” using haptic interface of steering and pedal to realize good cooperative characteristics between the driver and the system by optimizing the assistance level while minimizing the interference with human driver operation.
Recently, Autonomous Emergency Braking System (AEBS) have been already introduced to the markets to realize road traffic zero fatalities, and humans positively evaluated the performance of such assistance system safety function. However, such assistance system reaches its limit when there are unexpected moving obstacles appearing suddenly from poor visibility area, such as a corner of blind intersection, and a space behind a parked vehicle. This research focuses on a parked vehicle overtaking scenario which an occluded pedestrian suddenly might intend to cross a road, and risk predictive shared deceleration control system which can assist in avoiding potential collision risk is proposed. By conducting the driving simulator experiment, we investigated its functionality and the effectiveness of the proposed pre-braking intervention support.
This research focuses on the intersection right-turn scenario where an object darts out from blind spot of a congested oncoming traffic at the intersection. In such scenario, due to physical friction limit, collision may not be avoided by emergency braking of a driver or an active safety function such as Autonomous Emergency Braking System (AEBS). For improving safety under the limited physical friction potential, the main objective of this research is to develop a risk predictive right-turn assistance system by predicting the potential oncoming object to reduce the vehicle velocity in advance. The blind corner can be detected by on-board sensors without requiring information from infrastructure. The intersection right-turn assistance system is developed and it is evaluated that the system effectively decelerates the ego vehicle to a safe velocity that enables to avoid conflict by AEBS in the emergency cases.
The near-crash events relevant to vulnerable road users can lead to serious accidents. This research proposes a potential risk estimation method for predicting the criticality of driving situations with respect to driving context and driving behavior state. Near-miss incident events were extracted to conduct human error analysis as well as cause-and–effect chain studies of accidents, and the annotation parameters that describe the driving context were investigated to find their influence on the criticality of the recorded incidents. The obtained results can be used to develop next generation ADAS or to improve algorithms for autonomous driving for increasing their safety as well as the driver acceptance.
Advanced driver assistance systems and autonomous driving systems are being enhanced to deal with various types of collision use-cases. To handle those complicated scenarios, the rational and understandable integrated vehicle control algorithm on a 2D planar motion is required. Here, a new 2D motion control methodology assuming virtual repulsive force generated from obstacles is introduced. The use-case functions are the lane departure prevention vehicle control and the forward obstacle avoidance. The simulation is conducted in various driving situations not only for single lane change for the forward obstacle avoidance but also for the case of arbitrary approach angle with respect to direction towards obstacles. Assuming virtual repulsive force field, the collision avoidance path can be calculated as the combination of two parabolic curves and the control activation points also can be calculated analytically.
COPYRIGHT ©2019 PONGSATHORN LAB. ALL RIGHTS RESERVED.