Neural network learning control of robot manipulators using gradually increasing task difficulty

By Terence D. Sanger. 1994

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Continual Lifelong Learning with Neural Networks: A Review (2018, methodology): “This has inspired similar approaches in robotics (Sanger, 1994) and more recent machine learning methods studying the effects of curriculum learning in the performance of learning (Bengio, Louradour, Collobert, & Weston, 2009; Graves et al.”
From Baby Steps to Leapfrog: How “Less is More” in Unsupervised Dependency Parsing (2010, background): “Ideas that gradually make a task more difficult have been explored in robotics (typically, for navigation), with reinforcement learning (Singh, 1992; Sanger, 1994; Saksida et al., 1997; Dorigo and Colombetti, 1998; Savage, 1998; Savage, 2001).”
Curriculum learning (2009, result): “Similar ideas were also explored in robotics (Sanger, 1994), by gradually making the learning task more difficult.”
An Adaptive Backstepping Nonsingular Fast Terminal Sliding Mode Control for Robust Fault Tolerant Control of Robot Manipulators (2019, background): “In the initial approaches, PID or PD controller [15-16], computed torque controller (CTC) [6], intelligent and learning control [17-20], optimal control [21-22] and robust control [23-25], etc.”
A Survey on Curriculum Learning (2021, background): “Later related work [95], [98] in RL and robotics domains also discussed how to organize the presenting order of tasks from easy to hard.”
Neural Networks and Deep Learning (2018, unknown): “A number of methods were proposed in the early years [112, 422, 464] that showed the advantages of curriculum learning.”
A Review of Robot Learning for Manipulation: Challenges, Representations, and Algorithms (2019, background): “Such methods often utilize curricula provided by human designers [28, 322, 300], though several recent methods seek to automatically generate a curriculum [273, 374, 115].”
What Can Transformers Learn In-Context? A Case Study of Simple Function Classes (2022, methodology): “We exploit this by training our model using a curriculum [5, 20, 60, 73], where we train on a simpler distribution of functions in the beginning (e.g., linear functions with weight vectors restricted to a low-dimensional subspace) and gradually increase the function complexity.”
Short- and long-term changes in joint co-contraction associated with motor learning as revealed from surface EMG. (2002, background): “In robot J Neurophysiol • VOL 88 • AUGUST 2002 • www.jn.org learning, for example, higher learning rates were obtained by gradually increasing the speed or decreasing the stiffness as the learning progressed (Katayama et al. 1998; Sanger 1994).”
Curriculum Learning for Reinforcement Learning Domains: A Framework and Survey (2020, background): “The idea of using such curricula to train artificial agents dates back to the early 1990s, where the first known applications were to grammar learning (Elman, 1993; Rohde and Plaut, 1999), robotics control problems (Sanger, 1994), and classification problems (Bengio et al., 2009).”