| Research Abstract |
In this research, when a human operates a slave robot at a remote site indirectly by a master robot at an operation site, what kind of software required at operation and remote sites are investigated in this paper. For this purpose, we design intelligent mobile robots and robotic manipulators, and then we consider how a human being operates these robots with the help of several kinds of software.
First of all, we install on-line and off-line obstacle avoiding algorithms in mobile robots and robotic manipulators. Then, we evaluate them by convergence speed of a robot to its destination, or convergence robustness of a robot near its destination under sensing errors including GPS (Global Positioning System). Especially, we investigate which algorithm is suitable for supervising a non-holonomic mobile robot.
Secondly, when a human operates a robotic manipulator, what kind of space (Euclidean space or configuration space) is suitable for supervising the manipulator. Unfortunately because of f
… More
ew experiments, we cannot determine which space is better, but in a kind of task or a part in a task, one space is completely better than another space. Therefore, we will determine which is better in an arbitrary work or its part in future. Here, in order to construct a configuration space (obstacle), we propose a fast collision-check algorithm between links of a manipulator or a manipulator and its static obstacles. Also, we extend the algorithm to generate friction and impulse forces artificially in a 3-D graphics (virtual) world.
Finally, we build a master-slave robot system practically in this research, and then check usefulness of the shared autonomy system where intelligent robots and peoples run in the same environment. The system always includes a transmission delay between operation and remote sites, and a set of errors of robot, control, and sensing. Firstly, if master and slave robots are directly connected by a classic bilateral control, we check whether a human being at an operation site assembles many tasks while feeling friction and impulse forces made at a remote site comfortably. Secondly, since operation and remote sites far from each other, we prepare a graphics environment beside an operator, and the operator selects a good sequence of task motions while watching virtual slave robot and its environment. Here, we check flexibility of the remote operation experimentally if a predictive display of a slave site besides an operator exists at an operation site. In addition, we check comfortableness of the remote operation experimentally if an operator can feel friction and impulse forces artificially at a graphics computer (virtual environment). Finally, we check whether a slave robot is completely protected by the sensor-based path-planning (avoiding an unexpected collision between a robot and its obstacle). This is a fail-safe function of a slave robot at a remote site while overcoming robot, control, and sensing errors, and a large transmission delay. Less
|
