Search

This research proposes a new image visualization method of bulldozer and surrounding dangerous area to operator. Bulldozers are operated on uneven terrain which contains dangerous slope or obstacle. The objective of this research is to visualize the pose of bulldozer and surrounding terrain shape and to emphasis existence of dangerous terrain shape. Our proposed method consists of following three parts. First part is machine pose visualization. Machine pose is measured by on-board IMU and GNSS. The 3DCG model pose is changed according to measurement rotation and translation motion. Second part is terrain shape visualization. Terrain shape is measured by on-board laser radar and change projection plane shape. Third part is danger area detection and visualization. Danger area is detected by point density and incline of ground shape mesh. An experiment was conducted with a real bulldozer. The image which was generated by the proposed method was compared to what generated by the conventional method. Visibility of pose change of bulldozer and surrounding danger environment was improved.

Three-dimensional analysis of aurora is important for the research of solar wind and magnetic storms, because aurora reflects the relationship between solar wind and terrestrial magnetism. Therefore a method to reconstruct a three-dimensional shape of aurora precisely is demanded. In this paper, a method to measure three-dimensional shape of auroa by using only two fish-eye cameras is proposed. Two fish-eye cameras were installed at Fairbanks, Alaska, U.S.A to get the time series images of aurora. The images photographed by the cameras are performed calibration by using the star to estimate the camera attitude. The dense feature points are detected from the images of aurora by using template matching though the image of aurora has few characteristic pattern. Moreover, a feature point tracking between two images continuing in time is performed to improve precision of the feature point detection. Then the three-dimensional coordinates of feature points are calculated by triangulation, aurora shape is measured and visualized.

The hammering test has been widely used for inspection of social infrastructures because of its accuracy and efficiency of operation. In order to automate and apply the method to actual inspection sites, it is important for the system to calibrate itself against various environmental noise sources, such as strong winds, that influence the accuracy of the acoustic diagnosis. In this paper, a boosting based diagnostic algorithm with an updation rule for the weak learners that constitute the whole defect detector is proposed. Both clean and defective hammering sounds are templatized in the frequency domain, and inner parameters of the weak learners are optimized by using the pair of frequency template vectors. Furthermore, a new updation rule for template vectors is proposed. In experiments, our method was applied to defect detection of concrete test-pieces having slant cracks, which are considered to be serious defects in inspection sites. From the results, the effectiveness of the proposed method was confirmed.

In many sports, essential skills exist in terms of muscle activation in order to improve performance. However, it is difficult to identify the important muscle activation, and moreover novices cannot observe and imitate it since muscle activation is hardly visible. This study especially focuses on human rowing motion to extract an important skill in terms of muscle activation and we develop skill education system for novices to learn the skill. Firstly, strength and activation timing of lower leg muscle are evaluated to detect the expert skill based on their peak value and cross correlation during movement. Next, new skill education system is suggested by simultaneous presentation of real human movement and muscle activations of the targeted muscle. In the proposed system, muscle color is continuously changed based on its activation level for novices to perceive their muscle activation visually. Both experts and novices participated at our skill extraction experiment to compare their muscle activation. Results showed that experts had larger peak activation level and different profile in rectus femoris muscle. Therefore, our skill education system focused on rectus femoris muscle to present its activation level and movement simultaneously. In order to evaluate efficacy of the developed system, four different education instructions were conducted to novices; linguistic education, simultaneous presentation of motion and muscle activation of the expert, off-line presentation of novice's motion and muscle activation, and real time presentation of novice's motion and muscle activation. Performance of novices was evaluated after each instruction based on peak activation level, cross correlation, and exerted power on the rowing ergometer. These performance change after each education implied providing integrated view of novice's motion and muscle activation in real-time could improve performance of rowing motion.

This paper describes a development of bird's-eye view system for teleoperation. Emergent disaster recoveries are important when natural disaters occur. Unmanned construction is used to recover from natural disasters such as landslide and pyroclastic flow. This construction method has no risk of secondary disaster because workers teleoperate construction machines at remote site. However, traditional unmanned construction is an inefficient way because captured images from a camera mounted on a construction machine usually has blind spots. We propose an easy calibration method to create bird's-eye view images which can reduce blind spots from multiple cameras. Our method helps workers understand environment easily and operate construction machines safely. We evaluate working efficiency by integrating our bird's-eye view system into actual unmanned constuction systems under actual field conditions.

In daily life, there are many bottlenecks such as narrow passages, gaps between objects and entrances to rooms. At the bottlenecks, some deadlocks must occur so that mobile robots cannot move to their goals safely and efficiently. In order to solve this problem, it is necessary for robots working in daily life to give way to the others as humans do. The purpose of this research is to propose a new model for the mobile robot navigation in which the pedestrian intention is modeled with the probabilistic approach and show how important the give-way behavior is for the mobile robot navigation. In this research, we extend the Social Force Model with the intention vector to represent the intention change of the pedestrian and develop a new method to estimate others' intention. By modeling the pedestrian intention with probabilistic appoach, the proposed model solve the deadlocks at the bottleneck and realize the safe and efficient move of mobile robots.