Motion Estimation and Path Planning for Assistive Robotic Devices

Abstract

Abstract Assistive robotic devices have recently become a popular tool in various healthcare applications. To better assist users in their daily activities with robotic devices, adequate moving paths of joints need to be adopted based on user’s motions. In this paper, a motion predicting model was proposed. With the model developed using convolutional neural networks (CNNs), the corresponding type of motions can be determined efficiently in the initial state. A deriving procedure of common trajectories of desired motions has also been proposed using the approach of temporal alignment. These derived common trajectories are stored as a library. After the type of a specific motion being identified, paths are then synthesized to drive robotic devices with these derived common trajectories.