Neural Network-Based Robot Trajectory Generation

Daniel J. Simon

doi:10.1109/ICNN.1993.298615

Neural Network-Based Robot Trajectory Generation

Research output: Contribution to journal › Article › peer-review

Abstract

Interpolation of minimum jerk robot joint trajectories through an arbitrary number of knots is realized using a hardwired neural network. The resultant trajectories are numerical rather than analytic functions of time. This application formulates the interpolation problem as a contrained quadratic minimization problem over a continuous joint angle domain and a discrete time domain. Time is discretized according to the robot controller rate. The neuron outputs define the joint angles. An annealing-type method is used to prevent the network from getting stuck in a local minimum. The optimizing neural network and its application to robot path planning are discussed, some simulation results are presented, and the neural network method is compared with other minimum jerk trajectory planning methods

Original language	American English
Journal	IEEE International Conference on Neural Networks
Volume	1
DOIs	https://doi.org/10.1109/ICNN.1993.298615
State	Published - Mar 1 1993

Keywords

Annealing-type method
Continuous joint angle domain
Contrained quadratic minimization problem
Discrete time domain
Hardwired neural network
Interpolation problem
Minimum jerk robot joint trajectories
Robot controller rate
Robot path planning

Disciplines

Electrical and Computer Engineering
Robotics

Access to Document

10.1109/ICNN.1993.298615License: CC BY

http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=298615

Cite this

@article{eb9223ff0b154af391956aad5dc1ab87,

title = "Neural Network-Based Robot Trajectory Generation",

abstract = " Interpolation of minimum jerk robot joint trajectories through an arbitrary number of knots is realized using a hardwired neural network. The resultant trajectories are numerical rather than analytic functions of time. This application formulates the interpolation problem as a contrained quadratic minimization problem over a continuous joint angle domain and a discrete time domain. Time is discretized according to the robot controller rate. The neuron outputs define the joint angles. An annealing-type method is used to prevent the network from getting stuck in a local minimum. The optimizing neural network and its application to robot path planning are discussed, some simulation results are presented, and the neural network method is compared with other minimum jerk trajectory planning methods",

keywords = "Annealing-type method, Continuous joint angle domain, Contrained quadratic minimization problem, Discrete time domain, Hardwired neural network, Interpolation problem, Minimum jerk robot joint trajectories, Robot controller rate, Robot path planning",

author = "Simon, \{Daniel J.\}",

note = "D. Simon. (1993). Neural Network-Based Robot Trajectory Generation. IEEE International Conference on Neural Networks, 1, 540-545, doi: 10.1109/ICNN.1993.298615.",

year = "1993",

month = mar,

day = "1",

doi = "10.1109/ICNN.1993.298615",

language = "American English",

volume = "1",

journal = "IEEE International Conference on Neural Networks",

}

TY - JOUR

T1 - Neural Network-Based Robot Trajectory Generation

AU - Simon, Daniel J.

N1 - D. Simon. (1993). Neural Network-Based Robot Trajectory Generation. IEEE International Conference on Neural Networks, 1, 540-545, doi: 10.1109/ICNN.1993.298615.

PY - 1993/3/1

Y1 - 1993/3/1

N2 - Interpolation of minimum jerk robot joint trajectories through an arbitrary number of knots is realized using a hardwired neural network. The resultant trajectories are numerical rather than analytic functions of time. This application formulates the interpolation problem as a contrained quadratic minimization problem over a continuous joint angle domain and a discrete time domain. Time is discretized according to the robot controller rate. The neuron outputs define the joint angles. An annealing-type method is used to prevent the network from getting stuck in a local minimum. The optimizing neural network and its application to robot path planning are discussed, some simulation results are presented, and the neural network method is compared with other minimum jerk trajectory planning methods

AB - Interpolation of minimum jerk robot joint trajectories through an arbitrary number of knots is realized using a hardwired neural network. The resultant trajectories are numerical rather than analytic functions of time. This application formulates the interpolation problem as a contrained quadratic minimization problem over a continuous joint angle domain and a discrete time domain. Time is discretized according to the robot controller rate. The neuron outputs define the joint angles. An annealing-type method is used to prevent the network from getting stuck in a local minimum. The optimizing neural network and its application to robot path planning are discussed, some simulation results are presented, and the neural network method is compared with other minimum jerk trajectory planning methods

KW - Annealing-type method

KW - Continuous joint angle domain

KW - Contrained quadratic minimization problem

KW - Discrete time domain

KW - Hardwired neural network

KW - Interpolation problem

KW - Minimum jerk robot joint trajectories

KW - Robot controller rate

KW - Robot path planning

UR - https://engagedscholarship.csuohio.edu/enece_facpub/176

UR - http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=298615

U2 - 10.1109/ICNN.1993.298615

DO - 10.1109/ICNN.1993.298615

M3 - Article

VL - 1

JO - IEEE International Conference on Neural Networks

JF - IEEE International Conference on Neural Networks

ER -

Neural Network-Based Robot Trajectory Generation

Abstract

Keywords

Disciplines

Access to Document

Other files and links

Cite this