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

Abstract

Many works in robot teaching either focus on teaching a high-level abstract knowledge such as task constraints, or low-level concrete knowledge such as the motion for accomplishing a task. However, we show that both high-level and low-level knowledge is required for teaching a complex task sequence such as opening and holding a fridge with one arm while reaching inside with the other. In this paper, we propose a body role division approach, which maps both high-level task goals and low-level motion obtained through human demonstration, to robots of various configurations. 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 achieving a demonstrated motion, and body parts that are substitutional for adjusting the motion to achieve an instructed task goal. Our results show that our method scales to robots of different number of arm links, and that both high and low level knowledge is mapped to achieve a multi-step dual arm manipulation task. In addition, our results indicate that when either the high or low level knowledge of the task is missing, or when mapping is done without the role division, a robot fails to open a fridge door or is not able to navigate its footprint appropriately for an upcoming task. We show that such results not only apply to human-shaped robots with two link arms, but to robots with less degrees of freedom such as a one link armed robot.

Semantic constraints to represent common sense required in household actions for multi-modal Learning-from-observation robot

“…In a typical LfO system, a task is defined as the transition of a target object; for example, a contact state between polyhedral objects for part assembly by [1] or a topology of a string for tying a knot by [2]. To extend the LfO to household actions, [8] recently reported an LfO system that supports motion constraints derived from the linkage mechanism and the task representations were mapped on robots of various configuration by [9]. Although these previous systems achieved success in specific task domains, the tasks were defined with physical constraints; semantic constraints were ignored.…”

Hierarchical Reinforcement Learning Considering Stochastic Wind Disturbance for Power Line Maintenance Robot

“…for the local tightening policy, the robot needs to pay attention to the continuous feeding in the direction coaxial to the central axis of the bolt until tightening while meeting the collision conditions of the local approach sub-policy. In order to evaluate the coaxiality error between the end tool and the bolt, Equation (11) calculates the distance between the key points on the end tool and the straight line where the bolt is located, and takes the maximum of the two as the evaluation value. At the same time, considering the feed of the end tool along the central axis of the bolt, the position error for the end tool is used to characterize it.…”

“…It can also leverage the experience more effectively, resulting in significant performance gains. Sasabuchi and Wake et al [11] propose a body role division method suitable for different robots. It guides the configuration of the robot to complete a given task, which helps to teach a series of task sequences.…”

Task-grasping from human demonstration

“…One of the frameworks to address this difficulty is Learning-from-Observation (LfO), in which a human provides manipulation instruction to a robot through a oneshot demonstration [1], [2]. We have previously shown that the LfO framework can be used to teach the humanoid robot implicit high-level knowledge for the task, such as the semantic and physical constraints that exist between the environment and the grasping object, and the reaching body posture that is less likely to conflict with the environment [3]. Planning grasps based on these implicit knowledge has the potential to realize task-grasping.…”