ADAS (Advanced Driver Assistance Systems) requires not only real-time robust lane detection, both straight and curved, but also predicting upcoming steering direction by detecting the curvature of lanes. In this paper, a curvature lane detection algorithm is proposed to enhance the accuracy and detection rate based on using inverse perspective images and Gaussian Mixture Model (GMM) to segment the lanes from the background under various illumination condition. To increase the speed and accuracy of the lane detection, this paper used template matching, RANSAC and proposed post processing method. Through experiments, it is validated that the proposed algorithm can detect both straight and curved lanes as well as predicting the upcoming direction with 92.95% of detection accuracy and 50fps speed.

In the field of computer vision and robotics, bin picking is an important application area in which object pose estimation is necessary. Different approaches, such as 2D feature tracking and 3D surface reconstruction, have been introduced to estimate the object pose accurately. We propose a new approach where we can use both 2D image features and 3D surface information to identify the target object and estimate its pose accurately. First, we introduce a label detection technique using Maximally Stable Extremal Regions (MSERs) where the label detection results are used to identify the target objects separately. Then, the 2D image features on the detected label areas are utilized to generate 3D surface information. Finally, we calculate the 3D position and the orientation of the target objects using the information of the 3D surface.

This paper presents the implementation of altitude and attitude measurement algorithm using stereo camera for an unmanned aerial vehicle (UAV). Depth images are generated by calibrating the stereo cameras, and converted into 3D point cloud data. By applying a plane fitting algorithm to the resultant point cloud, altitude from ground level, and roll and pitch angles are extracted. To verify the performance, experimental results are provided by comparing with those of the motion caption system.

The accuracy of small and low-cost CCD cameras is insufficient to provide data for precisely tracking unmanned aerial vehicles (UAVs). This study shows how a quad rotor UAV can hover on a human targeted tracking object by using data from a CCD camera rather than imprecise GPS data. To realize this, quadcopter UAVs need to recognize their position and posture in known environments as well as unknown environments. Moreover, it is necessary for their localization to occur naturally. It is desirable for UAVs to estimate their position by solving uncertainty for quadcopter UAV hovering, as this is one of the most important problems. In this paper, we describe a method for determining the altitude of a quadcopter UAV using image information of a moving object like a walking human. This method combines the observed position from GPS sensors and the estimated position from images captured by a fixed camera to localize a UAV. Using the a priori known path of a quadcopter UAV in the world coordinates and a perspective camera model, we derive the geometric constraint equations that represent the relation between image frame coordinates for a moving object and the estimated quadcopter UAV's altitude. Since the equations are based on the geometric constraint equation, measurement error may exist all the time. The proposed method utilizes the error between the observed and estimated image coordinates to localize the quadcopter UAV. The Kalman filter scheme is applied for this method. Its performance is verified by a computer simulation and experiments.

This paper describes the design of an ankle rehabilitation robot for the force measurement of a severe stroke patient staying in a bed ward. The developed ankle rehabilitation robot was attached to a three-axis force/torque sensor that could detect force Fx, Fz, and torque Tz and measure the ankle rotation force (Fx) exerted on the ankle and the signal force Fz and torque Tz to be used as a safety device. The robot was designed and manufactured for bedridden stroke patients, and the robot program was manufactured to perform the flexibility rehabilitation exercise for ankle bending and to measure the ankle force to judge the degree of rehabilitation. According to the result of the characteristics test of the developed rehabilitation robot, it was safely operated while the ankle-bending flexibility rehabilitation exercise and the emergency situation were performed. Therefore, it is thought that the developed rehabilitation robot can be used for severe stroke patients.

This paper presents the neuro-fuzzy control method for balancing a two-wheel mobile robot. A two-wheel mobile robot is built for the experimental studies. On-line learning algorithm based on the back-propagation(BP) method is derived for the Takagi-Sugeno(T-S) neuro-fuzzy controller. The modified error is proposed to learn the B-P algorithm for the balancing control of a two-wheel mobile robot. The T-S controller is implemented on a DSP chip. Experimental studies of the balancing control performance are conducted. Balancing control performances with disturbance are also conducted and results are evaluated.

We propose a new type of semi-global matching (SGM) in order to solve a streaking problem arising from conventional SGM. Conventional SGM imposes a penalty to a pixel when the disparity of the pixel differs from that of the previous pixel along a scan path, and thus, disparity changes are not easily allowed, causing the streaking effect. The road surface is an appropriate target for such an effect, because the colors of the surfaces are very similar, and the image pixels corresponding to the surfaces show disparities that change very smoothly along the viewing direction. In contrast to conventional SGM, the new type of SGM imposes penalties depending on the disparity gradients, and thus, the streaking effect is controlled. The experimental results show the effectiveness of the proposed SGM method.

Various jumping robot platforms have been developed to carry out missions such as rescues, explorations, or inspections of dangerous environments. We suggested a jumping robot platform using energy conversion of the elastic body like the bar of a pole vault, which is the main part in which elastic force occurs. The compliant link was optimized by an optimization method based on Taguchi methodology, and the robot's leaping ability was improved. Among the parameters, the length, width, and thickness of the link were selected as design variables first while the others were fixed. The level of the design variables was settled, and an orthogonal array about its combination was made. In the experiment, dynamic simulations were conducted using the DAFUL program, and response table and sensitivity analyses were performed. We found optimized values through a level average analysis and sensitivity analysis. As a result, the maximum leaping height of the optimized robot increased by more than 6.2% compared to the initial one, and these data will be used to design a new robot.

As modern electronic devices get smaller and smaller, high-resolution, large Field-Of-View (FOV), fast, and cost-effective 3-dimensional (3-D) measurement is requested more and more. In particular, defect inspection machines using machine-vision technology nowadays require 3-D inspection as well as the conventional 2-D inspection. Phase Measuring Profilometry (PMP) is one of the fast non-contact 3-D shape measuring methods currently being extensively investigated in the electronic component manufacturing industry. The PMP system is well known and is successfully applied to measuring complex surface profiles with varying reflectance properties. However, for highly reflective surfaces, such as Ball Grid Arrays (BGAs), it has difficulty accurately measuring 3-D shapes. In this paper, we propose a new fast optical system that can eliminate the highly reflective saturated regions in BGA ball images. This is achieved by utilizing four Low Intensity Grating (LIG) images together with the conventional High Intensity Grating (HIG) images. Extensive experiments using BGA samples show a repeatability of under ${\pm}20um$ in standard deviation, which is suitable for most 3-D shape measurements of BGAs.

An Inertial Measurement Unit (IMU)-based Attitude and Heading Reference System (AHRS) can calculate attitude and heading information with long-term accuracy and stability by combining gyro, accelerometer, and magnetic compass signals. Motivated by this characteristic of the AHRS, this paper presents a Motion-Tracking and Localization (MTL) method for a person or walking robot using multi-AHRSs. Five AHRSs are attached to the two calves, two thighs, and waist of a person/walking robot. Joints, links, and coordinate frames are defined on the body. The outputs of the AHRSs are integrated with link data. In addition, a supporting foot is distinguished from a moving foot. With this information, the locations of the joints on the local coordinate frame are calculated. The experimental results show that the presented MTL method can track the motion of and localize a person/walking robot with long-term accuracy in an infra-less environment.