The level set method (LSM) has attracted much attention as a method that realizes a topology free active contour modeling. This method utilizes an implicit representation of a contour to be tracked, and is able to handle the topological change of the contour intrinsically. Various applications including motion tracking and 3D geometrical modeling have been proposed so far. However, the calculation cost of reinitialization and updating of the implicit function is considerably expensive as compared with conventional active contour models such as“Snakes”. To overcome this problem, we have proposed an efficient algorithm of the LSM named the Fast Level Set Method (FLSM) . This paper presents a 3D simultaneous motion tracking system of multiple targets using the FLSM, and shows a series of experimental results to verify the effectiveness of the system even in the event of occlusion. Characteristics of this system are as follows: (i) Topologically free, (ii) Robustness for self and mutual occlusion, (iii) Ease to calibrate several stereo cameras using ICP algorithm. In this paper, we present a robust 3D motion tracking system for target occlusion using the FLSM, and show some experimental results of tracking of multiple targets.
Dynamic gait generation of legged robots can considered as a problem to generate limit cycles in a phase space from the mathematical point of view. One of the basic necessary conditions of dynamic walking is to restore the mechanical energy that is dissipated by an impact between the swing leg and the floor at the transition instant. Energy-effective dynamic gait generation then yields the problem on how to restore the mechanical energy effectively during the swing phase. This paper first unifies the gait generation methods proposed so far by introducing a variable virtual gravity concept and two solution formulas. Taking up the energy tracking control and virtual passive dynamic walking as two typical methods, the paper analyzes the control performances from the view of points of energy efficiency and robust stability through numerical simulations. The obtained results are evaluated based on some criterion, and the conditions for maximum efficiency are theoretically clarified. Throughout this paper, we aim to clarify mathematical principle of energy-effective dynamic walking.
Weld beads, which exist on the corner of an automobile doorframe, must be grinded as if there are no weld beads on it. Grinding task is usually done by skilled workers. This weld bead has joint bumps and welding distortions around the joint area. These bumps and distortions have made it difficult to develop a practicable grinding robot system. We think that grinding task should create a smooth Free-Form surface around the weld bead area, not grinding off the weld beads only. In this paper, we propose a new grinding method, which consists of four distinctive techniques. The first technique is to measure the shape of the weld bead area using a high speed CMM (Coordinate Measuring Machine) . The second technique is to estimate a target Bezier surface based on the measurement data. The third technique focuses on calculating an optimum robot trajectory based on the grinding volume and the grinding belt properties. Finally, with the fourth technique, the robot will be precisely controlled to trace the calculated trajectory. This paper also presents the experimental results of grinding accuracy using this newly developed robot system.
A passive walker can walk down shallow slope with no energy source other than gravity. This motion is very attractive because its gait is really natural and ideal. Moreover, the walker can exhibit a stable limit cycle. Dynamics of passive walking is very interesting target and important for understanding human locomotion and developing the biped robots. Though the passive walkers are mechanically simple, they are a sort of hybrid systems with the switching condition which combines the nonlinear differential equations describing the swing motion and the leg-exchange. This makes it difficult to analyze. In this paper, we focus on the mechanism of stability of fixed points in passive walking. For the sake of simplicity and clarity as possible, we use a biped model known as the simplest walking model and treat the inter-leg angle at heel-strike as a variable. The equations of stability condition are derived from the eigenvalues of discrete dynamical system. We demonstrate a physical structure which forms the fixed points and a mechanism of its stability.
This paper describes the basic system and design of a quadruped walking robot TITAN XI. TITAN XI is designed to perform construction task on the operation site of steep slopes. Such a task is strongly demanded especially in Japan to construct highway in mountainous area. TITAN XI will be suspended by tethers on the steep slopes and walks around by four legs to avoid the damage of the ferro-concrete frames and perform the task to drill deep holes for rock bolts. We discussed fundamental design problems of TITAN XI; the selection of the system configuration, design of the leg mechanism and hydraulic driving system, suitable standard gait based on the intermittent crawl and the total controlling system. We have constructed mechanical model of TITAN XI 7, 000 [kg] in weight, and 3.7 [m] in leg length, and we have successfully demonstrated very basic walk of TITAN XI on even ground with the average velocity of 2 [cm/s] .
We acquire symbols based on the time sequence data obtained from our environments, and take the symbol manipulation for the data processing and generation, which is one of the higher functions of our intelligence. For the realization of the robot intelligence, it will be a fast way to design the symbol manipulation system. The scalability and spatio-temporal continuousness are important foci of the information processing system for the real world robot to process a large amount of information. In this paper, we design the spatially continuous symbol space using the dynamics based information processing system by putting the dynamics in the functional space. The utilization of the dynamics connects the continuous sensor signal and the discrete symbols, the allocation of the dynamics in the symbol space causes the spatially continuousness and the selection of the appropriate bases holds the scalability of the system.
This paper presents development of a preliminary model of a visible/thermal-infrared sensor for an on-orbit service robot. Visible and thermal-infrared sensors fused together function as super-eyes of the robot. As the result of the experiments, it is confirmed that diagnoses of the space instruments can be more efficiently carried out by fusion of visible and thermal-infrared image sensors. Furthermore, satellites or flying objects can be visualized by estimating their position and attitude from thermal-infrared images and can be displayed by superimposing visible images in the database on them, even if they are in eclipse and obscured by shade.
This paper describes 3-D shape modeling performed by a mobile robot. We propose a method in which the robot in the real environment acquires the accurate shape and constructs a 3-D model of an object using Computer Vision techniques under the condition that position and shape of the object is unknown. A 3-D model is reconstructed from image streams which are captured by a camera mounted on the robot. The experimental results show the effectiveness of our method.
In this paper, a simulator of environment and measurement that considers camera characteristics is developed for multiple autonomous mobile robots. RoboCup 4-legged robot league is chosen as the concrete target. The simulator introduces a sever/client system, and realizes separation of each robot's information, introduction of each robot's difference and distribution of processes. OpenGL is utilized to produce virtual images. So as to make a simulator reflect real environment, sensing noises should be considered. The simulator considers the effects of blur, element noises, vignetting, lens distortion and delayed exposure for each line of the CMOS image sensor. Experiments show the similarity between actual camera images and simulated camera images. Consequently, it is possible to verify programs for multiple autonomous mobile robots on the simulator, and therefore the proposed simulator is a useful tool for effective development of algorithms.
This research considers the mechanism design of the“Active Cord Mechanism (ACM) ”, which is a long and slender snake-type functional object that can perform three-dimensional movement. We designed a three dimensional ACM machine named ACM-R3 that can produce large output torque in spite of its small dimensions and light weight, and considered its three-dimensional steering control of two-dimensional serpentine locomotion. The meaning of steering in this research is the route change around the circumference of yaw, pitch and roll axes. Furthermore, novel propulsion forms, such as“Serpentine Locomotion with Lateral-Rolling” (whose control method is based on the steering of the roll axis) and“Sinus-Lifting Locomotion” (which is a special form of pitch axis steering) are examined. The latest model ACM-R3 was used to actually implement, for the first time ever, these new propulsion methods, and also to examine the motion ability experiemntally by measuring the torque produced at each node.
This paper reports the findings of a humanoid robot that pretends to listen to humans by effectively using its body properties in a route guidance situation. A human teaches a route to the robot, and the developed robot behaves as a human-like listener by utilizing both temporal and spatial cooperative behaviors to demonstrate that it is indeed listening to its human counterpart. The robot consists of many communicative units and rules for selecting appropriate units. A communicative unit realizes a particular cooperative behavior such as eye-contact and nodding, found through previous research. The rules for selecting communicative units were retrieved through WOZ experiments. An experiment was conducted to verify the effectiveness of the developed robot, and, as a result, the robot with cooperative behavior received higher subjective evaluation, which is rather similar to a human listener. The detailed analysis showed that this higher evaluation was due mainly to body movements as well as utterances. On the other hand, subjects' utterance to the robot was promoted by the robot's utterances but not by the body movements.
It is clarified throughout studies of passive dynamic walking mechanisms that the common necessary condition for dynamic gait generation comes from the requirement on mechanical energy restoration. Until now we have treated only rotational joints of the robot, whereas in this paper we considers a novel dynamic gait generation method based on mechanical energy restoration by parametric excitation phenomenon using telescopic leg actuation. We first introduce a simple biped walking model and a control law for telescopic leg motion, and analyze the typical walking pattern with respect to the change of control and physical parameters by numerical simulations. We then numerically analyze the energy efficiency in the presence of an elastic element by introducing criterions, and show that suitable adjustment of mechanical impedance dramatically increases the energy efficiency. Further, some extensions of the mechanism and control applications are investigated.