We are developing a vision system for an intelligent robot which can efficieently recognize many kinds of object. This system consists of two parts ; automatic generation of an object recognition procedure from a 3-D model (off-line part), feature extraction and object recognition by executing the procedure (on-line part) . This paper presents the off-line part of the system.
How to deal with the variations of object appearnces is the main issue in 3-D object recognition. In our approach, first, object appearances for various view directions are described with 2-D features which are derived from projections of the model. Then, the appearances for different view directions are analyzed by comparing descriptions for each direction, and a tree-like graph is constructed, where general descriptions visible in many directions are linked to more specific descriptions visible in fewer directions. Finally, the search order in the graph is determined by attaching a priority to each node of the graph. The capability of available image processing module in extracting features as well as the feature properties is considered in this priority evaluation
The generated tree-like graph is used as an object recognition procedure. The system searches through the graph in the order of the priority attached to each node and looks for the features in the description from an image.

View full abstract

In order to develop an advanced control system and a simulation system for a robot arm, it is necessary to establish an acculate identification method of a dynamical model.
The parameters of the dynamical model of the robot arm are composed of the link length, position of the center of mass, mass of link, inertia of link, viscous friction, and Coulomb friction. However, it is not necessary to estimate these parameters independently because if several combined constants appearing in the euqation of motion are obtained, then the control and the simulation can be done using only these combined constants.
In this paper, we firstly show that the viscous friction and Coulomb friction are obtained by measuring the relation between the current and angular velocity. Next we propose an effective identification method of inertial terms using the mechanical resonance occuring due to the stiffness of the joints.

View full abstract

A new-type sensor has been developed which possesses the functions of proximity and tactile. The sensor is attatched to a cylindrical arm surface of a manipulator in order to carry out both obstacle detection and interference checking in real time for collision avoidance control.
The paper introduces the structure of the detector part of the sensor at first. The means by which it detects obstacles is discussed next : A change of capacitance is used for proximity sensing, while a change of resistance of a pressure sensitive conductive rubber is used for tactile sensing. The changes on which the characteristics of the sensor is dependent are analyzed theoretically to show, with experimental values, that they are affected by the kinds and shapes of obstacles. Finally, the detecting circuit of the sensor which is composed mainly of an LC Hartley oscillator and the resultant characteristics of the sensor are shown.
The sensor has characteristics of having a high credibility in detecting the existence of any near-by and adjacent obstacle and making a quick response to their contacts. Furthermore, the detector part is elastic but yet sturdy and simple.

View full abstract

In order to reduce burdens of human operators in master-slave teleoperation of robots, it is necessary to introduce off-line teaching based on environment models and intelligent monitoring support for real tasks. This paper presents a new robot teleoperation method, which integrates interactive environment modeling, off-line teaching and task monitoring assisted by computer graphics on a novel display device, Multi-Media-Display (MMD) . MMD has been developed to accomplish “the real-time and stereo-scopic display of models superimposed on top of TV camera images”. With this facility, a sequence of typical teleoperation steps are carried out as follows. Those objects which play essential roles in a task are expected to have their shape models in advance. Their positions and orientations in the real environment can be determined by moving and rotating their graphics models interactively to fit on the TV camera images of corresponding real objects. Using these models, robot motions to achieve desired tasks can be taught in a graphics world. Even if the models are incomplete, the superimposed display enables visual collision checking between a manipulator and environmental objects through the motion simulation. In the real task monitoring, the superimposed images ease detection of errors, because it clearly presents the differences between expected images of models and real scenes. Throughout the job, MMD's real-time graphics and stereo-scopic images effectively improve the man-robot intef ace in 3 D space.

View full abstract

Collision avoidance problem of a robot manipulator whose workspace includes moving objects is considered in this paper. It is shown that the proposed personal computer simulation system can be used in a dialogue mode with a designer in order to check whether collision with obstacles is avoided or not and find out an appropriate movement.

View full abstract