小方 博之

We consider a teaching system that covers the stages of a task from planning to on-site teaching and propose for it a structured description based on task properties. A multimodal communication-teaching- advisor (MCTA) system is used for on-site teaching for a tool-manipulating task. The modular structure and object-oriented task handling program enable it to be applied to various robot systems and lets system developers customize the interface easily. The interface allows the operator and robot to share internal and external sensor data via nonverbal media. The intentions of an inexpert operator who has only common knowledge and ordinary abilities can then be transferred to the robot system via nonverbal operation. Experiments showed that two different types of robot controller could be controlled by this teaching system, and the nominal task program was designed for an environment with two controllers.

The development of robot systems must be simplified and made less expensive so that they can be used more widely in various industrial fields. We propose a component architecture of a robot-teaching system based on a universal task model, that enables fast and easy development of robot teaching systems for various types of work. This paper demonstrates the development of robot task design and management systems by applying this commponent architectures, and reports the development efficiency verified by simulation and experiments.

Modeling contact-state transitions is very troublesome. To deal with this problem, we have investigated a method of allocating local contact-state transition models (local models) to vertex pairs on which contact-state transitions occur and grouping the allocated local models according to a task process. In this paper, we propose a method of the allocation and the grouping from human demonstrations. Concretely, (i) for the allocation, vertex pairs where contact-state transitions occurred are found by using C-constrains in demonstrated data and (ii) the directions of contact-state transitions or the scope of the allocated local models are investigated for the grouping. We applied our method to some insertion tasks and tested it by performing the tasks by a compliance-controlled robot which had the generated groups of the local models. The experimental results showed the groups of local models work well as models of contact-state transition. As the result, we verified usefulness of our proposed method.

We propose a method for generating an assembly-task model from human demonstration. The model is pairs of operational coordinates and local contact-state transition models (local models) allocated in the coordinates. To generate the model, (1) the demonstrated data is segmented in each contact-state transition, (2) the segmented data is analyzed for setting operational coordinates using the component principal analysis, and (3) the local models are allocated to contact vertex pairs projected onto the operational coordinates. The usefulness of our method is discussed by the analysis results of demonstration in two assembly tasks and the robot task-execution using the generated models.

The development of robot systems must be simplified and made less expensive so that they can be used more widely in various industrial fields. We propose a component architecture of a robot-teaching system based on a universal task model, that enables fast and easy development of robot-teaching systems for various types of work.

Variety of design is required so that commodities will appeal to consumers. However, traditional mass production specializes in producing commodities with regular forms. This peeper proposes an assembly robot system that can produce the same products with various designs. Here, the robot learns a typical assembly task from the operator, and executes similar tasks in environments where the shape, scale, and location of objects may differ.

A new framework is proposed for generating trajectories for robots. This ''path planning by analogy'' approach uses problems and solutions acquired from previous operator teaching or automatic path planning. Unlike traditional path-planning methods, users need not specify goals or prepare search heuristics. Instead, users specify the correspondence of objects in the task environment to those in a reference environment. ''Path planning by analogy'' reduces the amount of operator instruction and the computational cost of making a path, and can generate a trajectory close to the desired trajectory.

Analysing character of object's motion under constraints is important to understand the mechanism of assembly tasks. Such analyses have been studied to solve the contact state transition problem. In those studies, the state of a manipulated object is often represented only by a combination of contacts between the manipulated object and objects in the environment. However, such a representation does not describe all the feature of motion in environments with concave or multiple objects. This paper points out some defects in traditional representation and suggest a solution addressing to the problem.

This paper introducesa method to extract a task knowledge from an example shown by an operator or generated by a planner, and to apply the knowledge to plan apath in similar environments. Our aim is to make efficient use of learning task resource and to plan a path in complex environments easily. Our method is based on the A<SUP>*</SUP> algorithm. We developed a technique to generate a suboptimal path with quite less amount of search nodes than the traditional A<SUP>*</SUP> algorithm, and to make a heuristic function that includes task knowledge. Examples are shown to verify the effect of our method.

A ''teaching by showing'' system using a graphic interface is proposed. This system provides 1) a user interface with which an inexperienced operator can easily teach a task, and 2) a task execution system that can operate the task in a different environment. The operator shows an example of assembly task movements in the virtual environmental created in the computer. A finite automaton defined task-dependently is used to interpret the movements as a sequence of high-level representations of operations. When the system is commanded to operate the learning task, the system observes the task environment, checks the geometrical feasibility of the task in the environment, and if necessary replans the sequence of operations so that the robot can complete the task. Then the system uses task-dependent interpretation rules to translate the sequence of operations into manipulator-level commands and executes the task by replicating the operator's movement in the virtual environment.

In the near future, mobile robots will play an important roll in various tasks in factory automations or indoor services. Such various tasks involve specifications like urgency or safety. As a result, planners must take those factors into account to generate motions for mobile robots. This paper proposes a trajectory planning method addressing this problem : Environment, moving obstacles, robot motion dynamics, and user's specifiaations are described with cost functions, and an optimization technique is applied. Our method can be applied to some time-varying environments that the time-space technique cannot treate, can consider the motion dynamics, and need not prepare particular algorithms for each user's specifiaction.

We present a method for teaching pick-and-place operation commands, generated from human assembly operation in a teaching environment, that allows a robot with mechanical constraints to perform an assembly task. The "teaching by showing" method, which is based on observing the operations of human workers, has recently been proposed for automatically generating robot commands. This method allows an operator with no knowledge of robots to teach movements to a robot. It is possible, however, that the difference between the degree of freedom of the human operator and of the robot, or differences between the teaching environment layout and the task environment layout may prevent the robot from completing the operation according to the taught sequence. In assembly tasks, the command execution order affects the completion of the assembly task. The method presented here represents operation priorities in the form of a graph, which is automatically generated from recognized human movement operations. When some operation cannot be performed by the robot, the operation sequence is reordered to complete the assembly task, on the condition that the graph remains the same. An example of a robot system is given and the results of a block assembly task are discussed.

Environments where a robot executes a task are not always the same as the environment where the robot learned the task. Traditional teaching playback robots execute a task by replaying the learning trajectory. Therefore such robots can only be used in environments where location and shape of objects are always the same. Furthermore, such robots are required to learn every movement from the biginning whenever the operator gives a new task. To reduce this inconvenience, a robust form of task description that is adaptable in different environments is needed. This paper discusses a method of describing a task robustly, and also presents algorithms (1) to interprete the learning motion to a task description, (2) to modify the description to the task environment, and (3) to generate the task motion from the modified description. The system architecture using this description is also discussed. The system consists of a teaching interface, a task interpreter, a sensor module, a task planner, and a manipulator module. We installed these modules in a simple system. This achieved a good performance in a block assembly task although it was taught in a different environment.