Robot Learning System Based on Adaptive Neural Control and Dynamic Movement Primitives

Yang, Chenguang; Chen, Chuize; He, Wei; Cui, Rongxin; Li, Zhijun

doi:10.1109/tnnls.2018.2852711

Cited by 285 publications

(145 citation statements)

References 31 publications

Supporting

Mentioning

143

Contrasting

Order By: Relevance

“…Then, the joint angular velocities can be computed based the difference of the joint angles. According to Yang et al (2018), we can assume that the joint angle of the initial position is zero. When the user's arm is moved from a pose T to a new pose P, the angle from pose T to P is the rotation angle.…”

Section: Calculation Of Joint Angular Velocity From Myo Armbandmentioning

confidence: 99%

“…In most cases (Billard et al, 2008;Yang et al, 2018;Fang et al, 2019), robots need to learn and execute many complex and repetitive tasks, which include learning the motion skills from observing humans performing these tasks, also known as teaching by demonstration (TbD). TbD is an efficient approach to reduce the complexity of teaching a robot to perform new tasks (Billard et al, 2008;Yang et al, 2018). With this approach, a human tutor demonstrates how to implement a task to a robot easily (Ewerton et al, 2019).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An Incremental Learning Framework to Enhance Teaching by Demonstration Based on Multimodal Sensor Fusion

Zhong

Yang

et al. 2020

Front. Neurorobot.

Self Cite

View full text Add to dashboard Cite

Though a robot can reproduce the demonstration trajectory from a human demonstrator by teleoperation, there is a certain error between the reproduced trajectory and the desired trajectory. To minimize this error, we propose a multimodal incremental learning framework based on a teleoperation strategy that can enable the robot to reproduce the demonstration task accurately. The multimodal demonstration data are collected from two different kinds of sensors in the demonstration phase. Then, the Kalman filter (KF) and dynamic time warping (DTW) algorithms are used to preprocessing the data for the multiple sensor signals. The KF algorithm is mainly used to fuse sensor data of different modalities, and the DTW algorithm is used to align the data in the same timeline. The preprocessed demonstration data are further trained and learned by the incremental learning network and sent to a Baxter robot for reproducing the task demonstrated by the human. Comparative experiments have been performed to verify the effectiveness of the proposed framework.

show abstract

Section: Calculation Of Joint Angular Velocity From Myo Armbandmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

An Incremental Learning Framework to Enhance Teaching by Demonstration Based on Multimodal Sensor Fusion

Zhong

Yang

et al. 2020

Front. Neurorobot.

Self Cite

View full text Add to dashboard Cite

show abstract

“…However, the influence of the number of Gaussian components on the model prediction accuracy has not been analyzed and considered. Besides, some other learning methods like conditional variational autoencoder (CVAE) [20], neural learning [21], experience graphs (E-Graphs) [22], and dynamic movement primitives (DMPs) [23] are also used in robot motion planning problem.…”

Section: Complexitymentioning

confidence: 99%

Robot Motion Planning Method Based on Incremental High‐Dimensional Mixture Probabilistic Model

et al. 2018

View full text Add to dashboard Cite

The sampling-based motion planner is the mainstream method to solve the motion planning problem in high-dimensional space. In the process of exploring robot configuration space, this type of algorithm needs to perform collision query on a large number of samples, which greatly limits their planning efficiency. Therefore, this paper uses machine learning methods to establish a probabilistic model of the obstacle region in configuration space by learning a large number of labeled samples. Based on this, the high-dimensional samples’ rapid collision query is realized. The influence of number of Gaussian components on the fitting accuracy is analyzed in detail, and a self-adaptive model training method based on Greedy expectation-maximization (EM) algorithm is proposed. At the same time, this method has the capability of online updating and can eliminate model fitting errors due to environmental changes. Finally, the model is combined with a variety of sampling-based motion planners and is validated in multiple sets of simulations and real world experiments. The results show that, compared with traditional methods, the proposed method has significantly improved the planning efficiency.

show abstract

“…Zeng et al combined the interactive force with the EMG signal to continuously predict the knee joint angle based on a state space model, and the Gaussian process was applied to improve the adaptivity of the method in practical applications [28]. Dynamic movement primitives (DMP)-based motion mode was also used to estimate the desired trajectory for pointto-point movement offline [29]. In general, robots with continuous human intention detection allow subjects to actively participate in the tasks rather than to be restricted to rigid predefined trajectory tracking, and show favorable results on motor function improvement [21].…”

Section: Introductionmentioning

confidence: 99%

Hybrid Active Control With Human Intention Detection of an Upper-Limb Cable-Driven Rehabilitation Robot

et al. 2020

View full text Add to dashboard Cite

Rehabilitation robots play an increasingly important role in the recovery of motor function for stroke. To ensure a natural physical human-robot interaction (pHRI) and enhance the active participation of subjects, it is necessary for the robots to understand the human intention and cooperate actively with humanlike characteristics. This study proposed a hybrid active control algorithm with human motion intention detection. The motion intention was defined as the desired position and velocity, which were continuously estimated according to the human upper-limb model and minimum jerk model, respectively. The motion intention was then fed into a hybrid force and position controller of an upper-limb cable driven rehabilitation robot (CDRR). And a three-dimensional reaching task without predefined trajectory was employed to validate the effectiveness of the proposed control algorithm. Experimental results showed that the control algorithm could continuously recognize the human motion intention and enabled the robot better movement performance indicated as smaller offset error, smoother trajectory, and lower impact. The proposed method could guarantee a natural pHRI and improve the engagement of the subjects, which has great potential in clinical applications.

show abstract

Robot Learning System Based on Adaptive Neural Control and Dynamic Movement Primitives

Cited by 285 publications

References 31 publications

An Incremental Learning Framework to Enhance Teaching by Demonstration Based on Multimodal Sensor Fusion

An Incremental Learning Framework to Enhance Teaching by Demonstration Based on Multimodal Sensor Fusion

Robot Motion Planning Method Based on Incremental High‐Dimensional Mixture Probabilistic Model

Hybrid Active Control With Human Intention Detection of an Upper-Limb Cable-Driven Rehabilitation Robot

Contact Info

Product

Resources

About