Gene

Quantum mechanics

Robot

Control (management)

Engineering

Control engineering

Exponential stability

Robustness (evolution)

Convergence (economics)

Economics

Biochemistry

Computer science

Physics

Nonlinear system

Mathematics

Mathematical proof

Mobile robot

Artificial intelligence

Geometry

Chemistry

Economic growth

Control theory (sociology)

Artificial neural network

IEEE Transactions on Neural Networks and Learning Systems

Various zeroing neural network (ZNN) models have been investigated to address the tracking control of robot manipulators for the capacity of parallel processing and nonlinearity handling. However, two limitations occur in the existing ZNN models. The first one is the convergence time that tends to be infinitely large. The second one is the research of robustness that remains in the analyses of stability and asymptotic convergence. To simultaneously enhance the convergence performance and robustness, this article proposes a new ZNN model by using a supertwisting (ST) algorithm, termed STZNN model, for the tracking control of mobile robot manipulators. The proposed STZNN model inherently possesses the advantages of finite-time convergence and robustness making the control process fast and robust. The bridge from the sliding mode control to the ZNN is built, and the essential connection between the ST algorithm and ZNN is explored by constructing a unified design process. Theorems and proofs about global stability, finite-time convergence, and robustness are provided. Finally, path-tracking applications, comparisons, and tests substantiate the effectiveness and superiority of the STZNN model for the tracking control handling of mobile robot manipulators.

A Novel Supertwisting Zeroing Neural Network With Application to Mobile Robot Manipulators