Learning Basis Representations of Inverse Dynamics Models for Real-Time Adaptive Control

2010 ◽  
pp. 668-675 ◽  
Author(s):  
Yasuhito Horiguchi ◽  
Takamitsu Matsubara ◽  
Masatsugu Kidode
Keyword(s):  
Adaptive Control ◽  
Real Time ◽  
Inverse Dynamics ◽  
Dynamics Models ◽  
Basis Representations

Real-Time Modular Deep Neural Network-Based Adaptive Control of Nonlinear Systems

2021 ◽  
pp. 1-1
Author(s):  
Duc M. Le ◽  
Max L. Greene ◽  
Wanjiku A. Makumi ◽  
Warren E. Dixon
Keyword(s):  
Neural Network ◽  
Adaptive Control ◽  
Nonlinear Systems ◽  
Real Time ◽  
Deep Neural Network

Multirate Integration for Real-Time Simulation of Wheel Loader Hydraulics

1999 ◽  
Author(s):  
Laurenţiu I. Buzdugan ◽  
Ole Balling ◽  
Peter Chien-Te Lee ◽  
Claus Balling ◽  
Jeffrey S. Freeman ◽  
...  
Keyword(s):  
Real Time ◽  
System Modeling ◽  
Pressure Fluctuations ◽  
Integration Scheme ◽  
Step Size ◽  
Integration Algorithm ◽  
Wheel Loader ◽  
Real Time Simulation ◽  
Time Simulation ◽  
Dynamics Models

Abstract This paper details a real-time simulation of an articulating wheel loader, which is comprised of a multibody system modeling the chassis and the bucket assembly and a set of subsystems. The hydraulic subsystem is modeled by a set of ODE’s which represent the oil pressure fluctuations in the system. An Adams-Bashforth-Moulton integration algorithm has been implemented using the Nordsieck form to develop a constant step-size multirate integration scheme, modeling the interaction between the hydraulic subsystem and multibody dynamics models. An example illustrating the simulation of a wheel loader bucket operation is presented at the end of the paper.

On the Inherent Characteristics of the Dynamics of Robot Manipulators

1991 ◽  
Author(s):  
Q. Tu ◽  
J. Rastegar
Keyword(s):  
Path Length ◽  
Degrees Of Freedom ◽  
Inverse Dynamics ◽  
Robot Manipulators ◽  
Control Point ◽  
Coordinate Space ◽  
Relative Significance ◽  
Harmonic Content ◽  
Trajectory Pattern ◽  
Dynamics Models

Abstract The inherent characteristics of the (nonlinear) dynamics of robot manipulators are studied. The study is based on a new method, referred to as the trajectory pattern method. The inverse dynamics models of the manipulator are divided into classes of inverse dynamics models, each corresponding to a different trajectory pattern. For each trajectory pattern, the structure of the resulting inverse dynamics model is fixed and is used to study the characteristics of the dynamics of the manipulator by examining the harmonic content of the required actuation torques (forces) and the relative significance of each harmonic. The harmonic content of the actuating torques is shown to be a function of the path length in the joint coordinate space and the harmonic content of the selected trajectory pattern, but is independent of the number of degrees-of-freedom of the manipulator. The relative contribution of each harmonic is a function of the path length, direction of motion, the position of the path of motion within the workspace of the manipulator, and the magnitude of the fundamental frequency. The study provides a systematic approach to path and trajectory planning from the vibration control point of view. As an example, the characteristics of the dynamics of a spatial 3R manipulator is studied for motions with two different path lengths, starting from a specified point and extending in different directions.

Real-time stable adaptive control implementation using a neural network processor

1995 ◽  
Author(s):  
Timothy Robinson ◽  
Mohammad Bodruzzaman ◽  
Kevin L. Priddy ◽  
Karl Mathia
Keyword(s):  
Neural Network ◽  
Adaptive Control ◽  
Real Time ◽  
Network Processor ◽  
Control Implementation

Inverse Dynamic Analysis and Linearisation of a High Speed Parallel Mechanism

2005 ◽  
Author(s):  
M. Necip Sahinkaya ◽  
Yanzhi Li
Keyword(s):  
Dynamic Analysis ◽  
Parallel Mechanism ◽  
High Speed ◽  
Inverse Dynamics ◽  
Motion Path ◽  
Model Parameters ◽  
Control Input ◽  
Inverse Dynamic ◽  
Inverse Dynamic Analysis ◽  
Dynamics Models

Inverse dynamic analysis of a three degree of freedom parallel mechanism driven by three electrical motors is carried out to study the effect of motion speed on the system dynamics and control input requirements. Availability of inverse dynamics models offer many advantages, but controllers based on real-time inverse dynamic simulations are not practical for many applications due to computational limitations. An off-line linearisation of system and error dynamics based on the inverse dynamic analysis is developed. It is shown that accurate linear models can be obtained even at high motion speeds eliminating the need to use computationally intensive inverse dynamics models. A point-to-point motion path for the mechanism platform is formulated by using a third order exponential function. It is shown that the linearised model parameters vary significantly at high motion speeds, hence it is necessary to use adaptive controllers for high performance.

A Novel Approach to Time Series Forecasting Using Model-Free Adaptive Control Framework

2020 ◽  
Author(s):  
Meenakshi Narayan ◽  
Ann Majewicz Fey
Keyword(s):  
Time Series ◽  
Adaptive Control ◽  
Real Time ◽  
Time Series Data ◽  
Mean Squared Error ◽  
Force Sensor ◽  
Sensor Data ◽  
Series Data ◽  
Model Free ◽  
Control Framework

Abstract Sensor data predictions could significantly improve the accuracy and effectiveness of modern control systems; however, existing machine learning and advanced statistical techniques to forecast time series data require significant computational resources which is not ideal for real-time applications. In this paper, we propose a novel forecasting technique called Compact Form Dynamic Linearization Model-Free Prediction (CFDL-MFP) which is derived from the existing model-free adaptive control framework. This approach enables near real-time forecasts of seconds-worth of time-series data due to its basis as an optimal control problem. The performance of the CFDL-MFP algorithm was evaluated using four real datasets including: force sensor readings from surgical needle, ECG measurements for heart rate, and atmospheric temperature and Nile water level recordings. On average, the forecast accuracy of CFDL-MFP was 28% better than the benchmark Autoregressive Integrated Moving Average (ARIMA) algorithm. The maximum computation time of CFDL-MFP was 49.1ms which was 170 times faster than ARIMA. Forecasts were best for deterministic data patterns, such as the ECG data, with a minimum average root mean squared error of (0.2±0.2).

Sign in / Sign up

Export Citation Format

Share Document