System Simulation for Robot Control Based on AI Approach

In this paper, we focus on real robot development and control for different surface conditions using an AI based approach. The core components of our existing robot are pressure sensor, servomotor and software-driven microcontroller. We performed robot system simulation for various ground surface conditions to control the robot with respect to pressure-sensing data that incorporates the two-way interactions between robot and ground. We have used an artificial neural network (ANN) approach for pressure-sensor-data analysis. The analysis result shows that above 25 hidden neurons and an increasing number of training cycles will deliver better performance in terms of mean square error (mse), learning time and improved nearest surface-pattern recognition. The analysis results are useful for next generation AI-chip development for real-time robot control and movement.