This paper describes robot language and software constructs of total intelligent robot, that is designed and implemented as general purpose research tool for robot system studies. The system is called COSMOS which is an abbreviation of Congnitive Sensor Motor Operations Study. It consists of high level robot language, arm and its control system, three dimensional vision, sensitive tactile sensor, and user interfaces. A host minicomputer and microcomputers are connected with the GPIB bus, and they communicate using primitive robot commands. The system consists hierachical control structure, and has modularity and extensibility in both hardware and software. For convenience of future extensions toward artificial intelligence technique, major parts of the system are implemeted in Lisp. Description of this paper includes robot language AL/ L (Assembly Language in Lisp) and it's environment model manager, structure of integrated robot software, and programming through tablet menu. First, design principle is described. Then, the language syntax of AL/ L and construction of it's language processor is explained. We recognize that an environment model manager plays an important role in the compilation of manipulation programs written in robot language. Through our development of teaching programs, it became clear that tablet is a good teaching device, and the system is effective for robot programming. This paper also discusses technical problems for future study.

View full abstract

This paper describes LARTS/T, a Language Aided Robotic Teleoperator System/Teaching oriented version. This system incorporates a high level robot language with a master-slave manipulator system. A new teaching-operating method which uses LARTS/T called LDMSM (Language Directed Master-Slave Manipulation) is also described.
The LARTS/T allows the operator to use language instruction and the master-slave operation cooperatively while preserving the merits of each. The most important feature of the LDMSM is that the structural contents of a task is written with a high level robot language and given to the system before-hand. In this way the system can learn spatial points needed for the task (environment) from the task performed by the operator following the movements derived from the structure. As a result of this feature, operating the system serves to teach to the system in the robot language.
In this paper, experiments performed on a prototype system are described. Results of these experiments demonstrate that the LDMSM is a flexible and intuitive teaching method for a teleoperator. They also show that the LARTS/T has the ability to learn environment and execute the tasks in modified ways while maintaining the immediate usability of teleoperators.

View full abstract

Detecting objects and measuring their positions and orientations are the most fundamental functions required to robot vision. RVL/A is a visual processing language for robots designed to use these functions easily. To obtain distance inf ormations and to process them in real-time, visual processing of RTV/A is executed on the range data to the object surface obtained by light-plane-intersecting method. The main f uatures of the language are as follows. (1) PVL/A provides the method to model the three-dmensional object shape using profiles of the object section. (2) Detection and measurement are executed by simple statements using instructions which refer to the model description. Experiments using a gasoline engine shows that RVL/A is applicable to programming of visual processing jobs in the hand-eye system.

View full abstract

The quadruped walking vehicle has a potential capability to be developed into a vehicle with high mobility and adaptability to the terrain by making use of its high degrees of motion freedom. The authors have been investigating on the gait control problems of the walking vehicle, such as the straightforward walking or crab walk of the vehicle on the rough terrain: The paper is to introduce a more generalized gait, namely a circular gait around turning axis arbitrary lacated, and the paper discusses a standard circular gait. The standard circular gait is the gait which maximize the speed of the walking and rotational angle in the circular walking, and the consideration thus forms a basis to discuss the advanced gait control problems. The paper formalizes the problems and analyses them by using mathematical optimization methods, such as non-linear programming. Computations are executed on TITAN III, the quadruped walking vehicle model constructed by the authors, and derived several characteristics of the optimum gait and final gait selection chart. The validity of these considerations are verified by the circular walking experiment of the TITAN III.

View full abstract

This paper proposes a new method of vehicle position and heading measurement for the automatic location and guidance of vehicles. The position and heading are determined by measuring the bearing angles of three points set on the vehicle with predetermined distances, and performing the trigonometric calculations by use of the angle data.
In order to demonstrate the above method, we present an on-line automatic position and heading measurement system consisting of a laser-beacon with constant rotational speed on a reference point and micro-processor with three laser-beam receivers on a vehicle.
The experiments of measuring position and heading of a standing vehicle using this system show good results. The standard deviation of the heading and the distance from the vehicle to the reference point are proportional to the distance. The best arrangement of the three points on a vehicle is the regular triangular. The position and heading errors due to movement of the vehicle are in inverse proportional to the rotational speed of the laser beacon and the distance from beacon to vehicle. From experiments and considerations on error, our system is effective and powerful for determining position and heading of vehicle.
This system has salient characteristics that positions and headings of many vehicles are simultaneously able to be measured by only one laser-beacon. Our system will be applied to monitoring and/or automatic guidance for passing ships in strait or harbor, running aircrafts on airport surfaces, vehicles in factory and farm, and so on.

View full abstract

In this paper, a new concept of a man-robot system (Advanced teleoperator) is proposed, in which, remotely operated tasks are efficiently performed. And key technologies that support the man-robot system are investigated. The advanced teleoperator can be regarded as a computer-aided teleoperator and/or a man-assisted robot, where the division of role for man and robot is essentially important. In the system, elaborate operations are performed by the robot, while the global task monitoring and high level decision-making are left to the man.
Key technologies for the advanced teleoperator include robot environment models, hybrid automatic/manual control of maipulators, man-interactive interpreters for task programs, and concurrent motion control. Several practical approaches are presented, in order to develop these elements.

View full abstract