Task-Oriented Motion Mapping on Robots of Various Configuration Using Body Role Division

Robotics and Automation Letters | , Vol 6(2): pp. 413-420

Publication

Many works in robot teaching either focus only on teaching task knowledge, such as geometric constraints, or motion knowledge, such as the motion for accomplishing a task. However, to effectively teach a complex task sequence to a robot, it is important to take advantage of both task and motion knowledge. The task knowledge provides the goals of each individual task within the sequence and reduces the number of required human demonstrations, whereas the motion knowledge contain the task-to-task constraints that would otherwise require expert knowledge to model the problem. In this letter, we propose a body role division approach that combines both types of knowledge using a single human demonstration. The method is inspired by facts on human body motion and uses a body structural analogy to decompose a robot’s body configuration into different roles: body parts that are dominant for imitating the human motion and body parts that are substitutional for adjusting the imitation with respect to the task knowledge. Our results show that our method scales to robots of different number of arm links, guides a robot’s configuration to one that achieves an upcoming task, and is potentially beneficial for teaching a range of task sequences.