This paper proposes helical rotation in-pipe locomotion inspired by spirochetes, a kind of bacteria. The principle of helical rotation enables us to develop a simple and small structure to propel inside a pipe with a flexible body actuated by shape memory alloy (SMA) wires. Such a robot is expected to move inside a winding tube without a special control strategy. Furthermore its rolling motion is unlikely to damage the inner wall of a pipe and its helical shape does not stuff a pipe. This paper analyzes the motion of the elastic body that is bent and rotated by SMA wires. Based on the analysis, we develop a prototype robot by developing SMA wires that can exert an appropriate force and that can also contract an appropriate amount of length. Experimental results show that the prototype robot can achieve helical rotating motion and propulsion in straight and curved tubes.
Generally, low backlash mechanisms are required for robot joints to perform tasks accurately. In this paper, we propose a new concept of a low backlash reducer using precessing crown gears. We call it the crown reducer. The features of this reducer are low backlash, high reduction ratio and ease of downsizing which we believe will contribute to the design of small mechanisms such as finger joints of robot hands. Firstly, the parts of the reducer suitable for downsizing are discussed. This will show that a layered construction of the nutation drive is most suitable. An explanation of the operation of the reducer and the method for generating the tooth profile of the stator gear that achieves multi-contact between rotor and stator gears will follow. And finally, evaluation of the performance of a prototype reducer will be discussed.
We proposed a global positioning technique in 3D environment using 3D geometrical map and a RGB-D camera based on a ND (Normal Distributions) voxel matching. Firstly, a 3D geometrical map represented by point-cloud is converted to ND voxels, and eigen ellipses are extracted. Meanwhile, ND voxels are also created from a range image captured by a RGB-D camera, and eigen ellipses and seven representative points are calculated in each ND voxel. For global localization, point-plane and plane-plane correspondences are tested and an optimum global position is determined using a particle filter. Experimental results show that the proposed technique is robust for the similarity in a 3D map and converges more stably than a standard maximum likelihood method using a beam model.
A posture control method for a hopping robot has been proposed in this paper. The control of the hopping robot is decomposed into two types of controls: the posture control that maintains the horizontal speed and the angles of a body and a leg, and the jumping control that keeps the jumping height. The posture motion of the hopping robot is approximated by a set of two linear continuous time systems for stance and flight modes, and two mode transition maps at touchdown and lift-off. We have shown that the posture of the robot can be stabilized by selecting a set of state feedback laws via numerically solving a minimax problem. Simulation results showed that the whole system is stabilized by the proposed control law, provided that the jumping heights maintained constant appropriately. We have also shown by simulation that the obtained feedback law provides smaller stabilizing inputs compared to one obtained by the Raibert's method. We have developed an actual one-legged hopping robot, and have shown that the robot can be stabilized and run at the specified speed by the proposed method and a simple jumping control.
This paper presents an unsupervised scene classification method for recognizing indoor scenes. Background and foreground features are respectively extracted using Gist and Scale-Invariant Feature Transform (SIFT) as feature representation based on context. Our method creates Bags of Features (BoF) to vote VWs (Visual Words) of SIFT and Gist features to a two-dimensional histogram. Moreover, our method can generate labels as a candidate of categories for input images while maintaining stability and plasticity together. Automatic labeling of category maps can be realized using labels created using Adaptive Resonance Theory (ART) as teaching signals for Counter Propagation Networks (CPNs). We evaluated classification accuracy of semantic categories such as a corridor and a room using KTH-IDOL datasets which are released for evaluating robot localization and navigation. The mean classification accuracy of Gist, SIFT, OC-SVM, PIRF, and our method reached to 39.7%, 58.0%, 56.0%, 63.6%, and 87.5%, respectively. The result of our method is 23.9% higher than that of PIRF. Moreover, we applied our method to a mobile robot for evaluating availability of our unsupervised classification method.
We proposed a new force sensing technique for estimation of gripping force, that detects extension of a driving component with two encoders and estimates applied force based on differences of two displacements. The technique fulfills three requirements for surgical robot; (i) high tolerance for noise, (ii) sterilizability and (iii) compactness in size. A method to compensate backrush and various effects of power transfer elements was proposed and verified by simulation and experiments. We could estimate gripping force with average errors of 0.32[N] for a surgical tool with one degree of freedom. We also developed gripping force control method based on the estimate force value, and implemented it on a master-slave surgical system. In the future, we will apply the technique to other joints of surgical tool.
We have proposed a Multi-portal Human Interface (M-HI) as an innovative human interface for power assist systems (PAS). The M-HI allows users to apply the operational force anywhere on the power assist system. Because of that, the effective workspace of the PAS is extended as their end-effecter can be controlled as a user desires. However, when users control the PAS on its intermediate joint, the control point motion is different from the end-effecter motion. This motion affects maneuverability of the PAS. In this paper, we consider the number of links and degrees of freedom on a multi-link manipulator. The relation between the control point motion and maneuverability is analyzed. The operational point motion harmonization (OPMH) is based on this relation. This method enables users to control the PAS with M-HI more accurately.