Teaching a sensor-enhanced robot under many task constraints is very difficult even for skilled operators, let alone novice operators. The problem is that the indexes for performing a robotic task, which the operators should simultaneously pay attention to, are not clear. To overcome this problem, we classify the task constraints into three indexes for robot, tool and sensor. We substitute the weighted indexes into the null-space method and generate robot motion for performing a task. We use this method in a tool-manipulation task such as welding and experimentally confirm the usefulness for reflecting the operator's teaching skill on the end-effector position and orientation using weights of indexes. Moreover we discuss the extraction of the operator's skills from teaching data as weights of these indexes.

A model of contact-state transitions is useful for motion planning, robot control, and sensing. Modeling the entire set of contact-state transitions every object, however, is very troublesome. To overcome the problem, we have developed a method for programming the model from human demonstration using a model of local contact-state transitions between a vertex pair (local model). Our method consists of three parts: (1) allocating the local model to vertex pairs where contact-state transitions will occur, (2) gathering them into groups according to an assembly sequence, and (3) unifying different groups resulting from different demonstrations of the same task. The allocation and grouping are done by simply showing a robotic system an assembly sequence through demonstration. The unification is done by comparing local models between different groups. We applied the method to two object-insertion tasks and did two kinds of experiments: (1) programming groups of local models from demonstration and (2) performing the demonstrated task by a robot using the programmed groups. The results showed the effectiveness in programming from demonstration and in the modeling problem.

A method is proposed for changing contact states during robotic assembly that is based on a multi-agent system. The agents that have its own model of the local contact state transition between corners of objects are prepared. Using its own model, each local agent individually senses the contact state, plans the motion for contact state transition, and suggests its motion to another agent. This agent integrates these suggestions and determines the robot motion for changing contact states. Without revising the model according to object shapes, our method determines the robot motion for changing contact states by changing the number of local agents. Simulation of single-peg and dual-peg insertion confirmed the effectiveness of this method.