Indexed by:
Abstract:
The manipulator must have a humanoid structure and human-like motion to achieve close and effective collaboration with humans. The anthropo-morphic manipulator has a structure similar to a human arm. We propose a human-like motion plan-ning method for moving target scenarios. We use dynamic movement primitives (DMPs) to model the anthropomorphic manipulator's end-effector motion. Compared to the conventional DMPs formalism, the proposed method contains the following extensions:1) an orientation-based spatial scaling,2) online temporal scaling parameter adaptation with moving targets feed-back. And we use a deep residual convolution neural network (ResCNN) to model the elbow motion of the anthropomorphic manipulator. Based on these models, real-time trajectory planning is performed for the moving target. In real-time motion planning, there is no need to model the moving target; only the moving tar-get's current position and velocity information is used to obtain a human-like motion of the anthropomorphic manipulator regarding posture and velocity. Simulation experiments on the Kinova Jaco2 in Coppeliasim show that our approach is practical. © 2023 IEEE.
Keyword:
Reprint Author's Address:
Email:
Source :
Year: 2023
Page: 1281-1286
Language: English
Cited Count:
SCOPUS Cited Count:
ESI Highly Cited Papers on the List: 0 Unfold All
WanFang Cited Count:
Chinese Cited Count:
30 Days PV: 5
Affiliated Colleges: