For autonomous mobility of unmanned or tele-operated vehicles, the visual navigation capabilities such as obstacle detection and terrain estimation are essential. This paper presents a robust model-based approach for detecting traversable regions and obstacles even in off-road environments. In our approach, a model-based polynocular stereo, which we call tilted-plane-based stereo, is utilized. We newly define the algorithm by using a group of slant parallel planes to be searched in the spatial disparity space. It is capable of directly measuring height differences in the view field. By checking abrupt jumps in height in the 3D space, traversable regions of the terrain can be effectively estimated. Then, the obstacles belong to the untraversable regions are robustly detected. We show successful results of the proposed method for different obstacles in the real scenes.
We have proposed a self-reconfigurable parallel robot, which can be configured to 4R and 5R closed kinematic chains. By mounting it on a crawler mechanism, this paper proposes a parallel mechanism mobile robot. The combined mobile robot can gain some useful functionalities from the advantage of its parallel mechanism other than just locomotion, such as carrying an object by making use of its shape and getting over a bump by control of its center of gravity or zmp. In general, for a crawler robot to get over a vertical bump, friction is necessary in the vertical direction of the bump. The proposed sequence of getting over a bump does not rely on friction. Furthermore, cooperation of two or more such robots gains functionalities such as forming three-dimensional structures. Using two robots, we verify that the 4R robot can elevate the 5R robot, which enables the latter to reach a certain height in which it cannot alone. We analyze the statics of this motion to evaluate the necessary joint torque of the 4R robot.
For communication robots, it is important to find a communication partner and attract his or her attention in our daily environment. In this paper, we propose a method for communication robots to find a human actively and attract attention in order to communicate with human. We apply Markov chain Monte Carlo Algorithms (MCMC) to human detecting and tracking behaviors with a humanoid robot that has 4 types of sensors. Thus, by utilizing our method, the robot can find and track human with irregular motion in complicated daily environment. While tracking human, it tries to attract attention by verbal and nonverbal communication. We verify the validity of our method by performing experiments with a humanoid type communication robot, named Robovie.
In rescue robot system, it is desirable to be carefully controlled by the operator because of the complexity of the environment and the variety of victim's condition. But it is very difficult for one operator to do the rescue works with processing all of sensory information at the same time. Therefore it is required to bring a feedback control system to utilize multi-sensory information into the rescue robot system. In this paper, the manual control and sensor feedback control by using multiple tactile sensors of rescue robot for the handling a part of human body is introduced.
This paper presents a load-sensitive continuously variable transmission (CVT) for finger joints. Fingers of robot hands require force when grasping an object and speed when opening and closing. Therefore a CVT is ideal to improve the power transmission of a finger joint. Existing friction CVTs are too big and heavy to be installed in a finger joint. By focusing on the fact that finger joints do not necessarily rotate 360 degrees, this paper presents a remarkably simple and small load-sensitive CVT consisting of a five-bar linkage and a torsion coiled spring. Experimental results show that the CVT can increase its reduction ratio from 0.5 to 3.3 in response to a load. We have developed a two-finger gripper with the CVTs, which can grasp an object powerfully and manipulate quickly. These motions would be impossible without the CVT.
In this paper, we propose a new index of manipulator's dynamic capability named Impedance Matching Ellipsoid, or IME, for serial link manipulators. Several indexes have been proposed in the past to illustrate statically and dynamically capability of a robot manipulator. For example, Dynamic Manipulability Ellipsoid (DME) describes a distribution of hand acceleration produced by normalized joint torque. Manipulating-Force Ellipsoid (MFE) denotes static force transmission from joints to a hand. On the other hand, the proposed IME illustrates dynamic torque-force transmission efficiency from actuators at joints to an object held at a hand. The concept of the IME involves a wide range of proposed indexes proposed as measures of manipulator's capability. The DME and MFE are both derived as a typical representation of the IME. This paper demonstrates the IME with numerical examples including optimal leg posture for a jump robot, optimum active stiffness control, and an extension to a free-flying manipulator.
I discuss the static properties of multi-DOF (Degrees Of Freedom) mechanisms in respect to‘power dimension’. Especially, in this paper, I describe basic formulas and criteria to analyze and evaluate the static output power properties. The power dimensional analysis is effective for improving fundamental performances of multi-DOF mechanisms. For example, improving output power, improving energy efficiency, reducing weight and so on. At the beginning of this paper, I derive basic formulas necessary for power dimensional analysis. Next, I describe the classification of‘joint power states’according to the‘sign of power’. Then I show the harmful effects of the states; some joints generate positive and others generate negative power simultaneously. I also show the criterion to evaluate the effects. Next, I focus on the relationships between ‘joint power states’ and ‘possible output power’. I derive the range of possible output power under the condition that each joint has limitation of power. And I show the conditions to maximize output power.
This paper describes an ego-motion estimation method by integrating multiple scan matching results. The method considers both the uncertainty of scan matching results and that of estimated ego-motions, and not only estimates the latest robot ego-motion but also updates previous ego-motions. The estimation process is formulated as an iterative one using Kalman filter. We implement the ego-motion estimation method using an omnidirectional stereo-based scan matching method which considers the uncertainty of the range data, and estimates the uncertainty of the result of the scan matching. Experimental results show the effectiveness of the proposed method.