An Adaptive Deep Belief Networkbased Intelligent moving Robot for Navigation Control using Mamdani-Sugeno Fuzzy Inference System

Abstract

The Intelligent Moving Robots (IMR) are designed to understand instructions and act accordingly in an independent manner. They use sensors that operate with the help of the Internet of Things(IoT) and Deep Learning (DL) for interpreting and navigating the directions following environmental conditions. Recent advancements use the Artificial Neural Network(ANN) and an Adaptive Neuro-Fuzzy Inference System(ANFIS) to model an efficient engineering system. In this work, a hybrid fuzzy inference system, MSFIS(Mamdani-Sugeno Fuzzy Inference System), is proposed along with Adaptive Deep Belief Networks(ADBN) for identifying and tracing the Direction Finding(DF) capability of the IMR. The MSFIS uses parameters in the range 4:4 (4 Input and 4 Output Parameters). The inputs used are Front View (FV), Left View(LV), Right View(RV), and Back View(BV), and the output (4 directions) might depend on the speed of the wheels used in the Robot. Four directions are used at the output for navigation purposes. The results obtained from simulating the experiments confirm that the suggested navigation controller demonstrates superior viability, efficiency, and resilience. Compared with the existing system, the proposed system outperforms well in accuracy and sensitivity, proving it is well efficient in navigating any new environment.

Keywords

• adaptive deep belief networks
• intelligent moving robots (IMR)
• internet of things (IoT)
• mamdani-sugeno fuzzy inference system