Learning over time using a neuromorphic adaptive control algorithm for robotic arms

  title={Learning over time using a neuromorphic adaptive control algorithm for robotic arms},
  author={Lazar Supic and Terrence C. Stewart},
In this paper, we explore the ability of a robot arm to learn the underlying operation space defined by the positions (x, y, z) that the arm’s end-effector can reach, including disturbances, by deploying and thoroughly evaluating a Spiking Neural Network SNN-based adaptive control algorithm. While traditional control algorithms for robotics have limitations in both adapting to new and dynamic environments, we show that the robot arm can learn the operational space and complete tasks faster over… 

Figures and Tables from this paper



Neuromorphic NEF-Based Inverse Kinematics and PID Control

A NEF-based neuromorphic algorithms for inverse kinematics and PID control are developed, which are used to manipulate 6 degrees of freedom robotic arm and offer high performing and energy-efficient neuromorphic control.

A spiking neural model of adaptive arm control

The model proposes novel neural computations within these areas to control a nonlinear three-link arm model that can adapt to unknown changes in arm dynamics and kinematic structure, and demonstrates the mathematical stability of both forms of adaptation.

Nengo and Low-Power AI Hardware for Robust, Embedded Neurorobotics

In this paper we demonstrate how the Nengo neural modeling and simulation libraries enable users to quickly develop robotic perception and action neural networks for simulation on neuromorphic

Closed-Loop Neuromorphic Benchmarks

This work presents a methodology for generating closed-loop benchmarks that makes use of a hybrid of real physical embodiment and a type of “minimal” simulation, and shows that an error-driven learning rule can consistently improve motor control performance across a randomly generated family ofclosed-loop simulations.

MuJoCo: A physics engine for model-based control

A new physics engine tailored to model-based control, based on the modern velocity-stepping approach which avoids the difficulties with spring-dampers, which can compute both forward and inverse dynamics.

Smooth Human-Robot Interaction in Robot-Assisted Rehabilitation

  • D. ErolN. Sarkar
  • Engineering
    2007 IEEE 10th International Conference on Rehabilitation Robotics
  • 2007
The goal of this work is to develop a control framework to provide robotic assistance for rehabilitation tasks to the subjects in such a manner that the interaction between the subject and the robot

A Solution to the Dynamics of the Prescribed Error Sensitivity Learning Rule

The discrete dynamical system for the case of constant inputs and no noise is solved to show that the decoding vectors given by the NEF have a simple closed-form expression in terms of the number of simulation timesteps.

Smooth and near time-optimal trajectory planning of industrial robots for online applications

A new smooth online near time‐optimal trajectory planning approach to reduce the consuming time compared to the conventional dynamics trajectory planning methods can avoid Range's phenomenon and also guarantee the continuity of torque.

PID control system analysis, design, and technology

It is seen that many PID variants have been developed in order to improve transient performance, but standardising and modularising PID control are desired, although challenging, and the inclusion of system identification and "intelligent" techniques in software based PID systems helps automate the entire design and tuning process to a useful degree.

Nengo: a Python tool for building large-scale functional brain models

Nengo 2.0 is described, which is implemented in Python and uses simple and extendable syntax, simulates a benchmark model on the scale of Spaun 50 times faster than Nengo 1.4, and has a flexible mechanism for collecting simulation results.