scholarly journals Predictive Control for Small Unmanned Ground Vehicles via a Multi-Dimensional Taylor Network

2022 ◽  
Vol 12 (2) ◽  
pp. 682
Author(s):  
Yuzhan Wu ◽  
Chenlong Li ◽  
Changshun Yuan ◽  
Meng Li ◽  
Hao Li
Keyword(s):  
Objective Function ◽  
Real Time ◽  
Predictive Control ◽  
Tracking Control ◽  
Learning Algorithm ◽  
Back Propagation ◽  
Unmanned Ground Vehicles ◽  
Ground Vehicles ◽  
Velocity Signal ◽  
The Stability

Tracking control of Small Unmanned Ground Vehicles (SUGVs) is easily affected by the nonlinearity and time-varying characteristics. An improved predictive control scheme based on the multi-dimensional Taylor network (MTN) is proposed for tracking control of SUGVs. First, a MTN model is used as a predictive model to construct a SUGV model and back propagation (BP) is taken as its learning algorithm. Second, the predictive control law is designed and the traditional objective function is improved to obtain a predictive objective function with a differential term. The optimal control quantity is given in real time through iterative optimization. Meanwhile, the stability of the closed-loop system is proved by the Lyapunov stability theorem. Finally, a tracking control experiment on the SUGV model is used to verify the effectiveness of the proposed scheme. For comparison, traditional MTN and Radial Basis Function (RBF) predictive control schemes are introduced. Moreover, a noise disturbance is considered. Experimental results show that the proposed scheme is effective, which ensures that the vehicle can quickly and accurately track the desired yaw velocity signal with good real-time, robustness, and convergence performance, and is superior to other comparison schemes.

scholarly journals An Adaptive RBF-NMPC Architecture for Trajectory Tracking Control of Underwater Vehicles

Machines ◽  
2021 ◽  
Vol 9 (5) ◽  
pp. 105
Author(s):  
Zhenzhong Chu ◽  
Da Wang ◽  
Fei Meng
Keyword(s):  
Model Predictive Control ◽  
Prediction Model ◽  
Real Time ◽  
Predictive Control ◽  
Trajectory Tracking ◽  
Tracking Control ◽  
Working Environment ◽  
Gray Wolf ◽  
Trajectory Tracking Control ◽  
System Output

An adaptive control algorithm based on the RBF neural network (RBFNN) and nonlinear model predictive control (NMPC) is discussed for underwater vehicle trajectory tracking control. Firstly, in the off-line phase, the improved adaptive Levenberg–Marquardt-error surface compensation (IALM-ESC) algorithm is used to establish the RBFNN prediction model. In the real-time control phase, using the characteristic that the system output will change with the external environment interference, the network parameters are adjusted by using the error between the system output and the network prediction output to adapt to the complex and uncertain working environment. This provides an accurate and real-time prediction model for model predictive control (MPC). For optimization, an improved adaptive gray wolf optimization (AGWO) algorithm is proposed to obtain the trajectory tracking control law. Finally, the tracking control performance of the proposed algorithm is verified by simulation. The simulation results show that the proposed RBF-NMPC can not only achieve the same level of real-time performance as the linear model predictive control (LMPC) but also has a superior anti-interference ability. Compared with LMPC, the tracking performance of RBF-NMPC is improved by at least 43% and 25% in the case of no interference and interference, respectively.

scholarly journals APPLICATION OF LEARNING REINFORCEMENT METHOD IN ROBOTIZED AND AUTOMATED FORESTRY SYSTEMS

2020 ◽  
Vol 10 (1) ◽  
pp. 256-265
Author(s):  
Andrey Tolstyh ◽  
D Stupnikov ◽  
Sergey Malyukov ◽  
Aleksandr Luk'yanov ◽  
Yuriy Lunev
Keyword(s):  
Physical Environment ◽  
Learning Algorithm ◽  
Back Propagation ◽  
Industrial Robots ◽  
Environment Model ◽  
Random Position ◽  
Learning Reinforcement ◽  
The Stability ◽  
Two Stages ◽  
The Cost

Abstract Currently, most large enterprises are actively using industrial robots and other automated solutions. This allows a significant increase in productivity and quality of work performed. This article gave a brief overview of modern industrial robots, their operating principle, basic components and systems. A reinforcement learning algorithm was developed and tested. The task of constructing a learning algorithm with reinforcement was divided into two stages: modeling the environment and description and optimization of the cost function. Since industrial robotic systems operate in the real world, the environment model should reflect basic physical laws. Therefore, the pyBullet library of the physical environment was chosen as the physical environment for testing. After modeling the manipulator in the selected physical medium, it was given the trivial task of touching a given object with the capture of the manipulator. An artificial neural network was used as an agent interacting with the environment. The inputs were the coordinates of the object and the existing angles of rotation of the articulated joints of the robot. Outputs - angle of rotation of joints at this step. This network was trained using the back propagation method, Adam modification. The system was trained for about 12 hours. Success is achieved in 95% of cases when testing the stability of the system (random position of the cylinder). In future, it is planned to test the obtained models on bench samples

Path planning and robust fuzzy output-feedback control for unmanned ground vehicles with obstacle avoidance

2020 ◽  
pp. 095440702097831
Author(s):  
Yimin Chen ◽  
Chuan Hu ◽  
Yechen Qin ◽  
Mingjun Li ◽  
Xiaolin Song
Keyword(s):  
Feedback Control ◽  
Path Planning ◽  
Obstacle Avoidance ◽  
Output Feedback ◽  
Output Feedback Control ◽  
Unmanned Ground Vehicles ◽  
Ground Vehicles ◽  
Velocity Signal ◽  
Moving Obstacles ◽  
Avoidance Strategy

Obstacle avoidance strategy is important to ensure the driving safety of unmanned ground vehicles. In the presence of static and moving obstacles, it is challenging for the unmanned ground vehicles to plan and track the collision-free paths. This paper proposes an obstacle avoidance strategy consists of the path planning and the robust fuzzy output-feedback control. A path planner is formed to generate the collision-free paths that avoid static and moving obstacles. The quintic polynomial curves are employed for path generation considering computational efficiency and ride comfort. Then, a robust fuzzy output-feedback controller is designed to track the planned paths. The Takagi–Sugeno (T–S) fuzzy modeling technique is utilized to handle the system variables when forming the vehicle dynamic model. The robust output-feedback control approach is used to track the planned paths without using the lateral velocity signal. The proposed obstacle avoidance strategy is validated in CarSim® simulations. The simulation results show the unmanned ground vehicle can avoid the static and moving obstacles by applying the designed path planning and robust fuzzy output-feedback control approaches.

The Role of Model Fidelity in Model Predictive Control Based Hazard Avoidance in Unmanned Ground Vehicles Using LIDAR Sensors

2013 ◽  
Author(s):  
Jiechao Liu ◽  
Paramsothy Jayakumar ◽  
James L. Overholt ◽  
Jeffrey L. Stein ◽  
Tulga Ersal
Keyword(s):  
Model Predictive Control ◽  
Predictive Control ◽  
Vehicle Dynamics ◽  
High Fidelity ◽  
Unmanned Ground Vehicles ◽  
Ground Vehicles ◽  
Model Fidelity ◽  
Hazard Avoidance ◽  
The Impact

Unmanned ground vehicles (UGVs) are gaining importance and finding increased utility in both military and commercial applications. Although earlier UGV platforms were typically exclusively small ground robots, recent efforts started targeting passenger vehicle and larger size platforms. Due to their size and speed, these platforms have significantly different dynamics than small robots, and therefore the existing hazard avoidance algorithms, which were developed for small robots, may not deliver the desired performance. The goal of this paper is to present the first steps towards a model predictive control (MPC) based hazard avoidance algorithm for large UGVs that accounts for the vehicle dynamics through high fidelity models and uses only local information about the environment as provided by the onboard sensors. Specifically, the paper presents the MPC formulation for hazard avoidance using a light detection and ranging (LIDAR) sensor and applies it to a case study to investigate the impact of model fidelity on the performance of the algorithm, where performance is measured mainly by the time to reach the target point. Towards this end, the case study compares a 2 degrees-of-freedom (DoF) vehicle dynamics representation to a 14 DoF representation as the model used in MPC. The results show that the 2 DoF model can perform comparable to the 14 DoF model if the safe steering range is established using the 14 DoF model rather than the 2 DoF model itself. The conclusion is that high fidelity models are needed to push autonomous vehicles to their limits to increase their performance, but simulating the high fidelity models online within the MPC may not be as critical as using them to establish the safe control input limits.

IMPLEMENTING BACK-PROPAGATION- THROUGH-TIME LEARNING ALGORITHM USING CELLULAR NEURAL NETWORKS

1999 ◽  
Vol 09 (06) ◽  
pp. 1041-1074 ◽  
Author(s):  
TAO YANG ◽  
LEON O. CHUA
Keyword(s):  
Real Time ◽  
Time Complexity ◽  
Large Scale ◽  
Learning Algorithm ◽  
State Of The Art ◽  
Back Propagation ◽  
Cellular Neural Network ◽  
Small Scale ◽  
Von Neumann ◽  
On Line

In a programmable (multistage) cellular neural network (CNN) structure, the CPU is a CNN universal chip which supports massively parallel computations on patterns and images, including videos. In this paper, we decompose the structure of a class of simultaneous recurrent networks (SRN) into a CNN program and run it on a von Neumann-like stored program CNN structure. To train the SRN, we map the back-propagation-through-time (BTT) learning algorithm into a sequence of CNN subroutines to achieve real-time performance via a CNN universal chip. By computing in parallel, the CNN universal chip can be programmed to implement in real time the BTT learning algorithm, which has a very high time complexity. An estimate of the time complexity of the BTT learning algorithm based on the CNN universal chip is presented. For small-scale problems, our simulation results show that a CNN implementation of the BTT learning algorithm for a two-dimensional SRN is at least 10,000 times faster than that based on state-of-the-art sequential workstations. For the few large-scale problems which we have so far simulated, the CNN implemented BTT learning algorithm maintained virtually the same time complexity with a learning time of a few seconds, while those implemented on state-of-the-art sequential workstations dramatically increased their time complexity, often requiring several days of running time. Several examples are presented to demonstrate how efficiently a CNN universal chip can speed up the learning algorithm for both off-line and on-line applications.

Sign in / Sign up

Export Citation Format

Share Document