M.K. Rashid and A.M. Al-Shabibi (Oman)
Mobile robot, gyrover, neurofuzzy control, ANFIS.
The stabilization of a single wheel mobile robot has attracted several studies in robotic area. However, the budget required for building experimental setups capable of investigating isolated parameters has generated concerns for implementing new simulation methods and techniques. In this work a simulation platform for testing different control tactics to stabilize a single wheel mobile robot has been developed. The graphical representation of the robot, the dynamic solution, and the control scheme are all integrated on a common computer platform using Visual Basic. Simulation indicates that by using manual operation we can build a control scheme for such robot without knowing details regarding its internal structure or dynamic behaviour. Twenty five rules are extracted and implemented using Takagi-Sugeno’s fuzzy controller with significant achievement in controlling robot motion during the dynamic simulation. The resulted data from the successful implementation of the fuzzy model are used to utilize and train a neurofuzzy controller using ANFIS scheme to produce further improvement in robot performance. MOTOR Figure 1 The Basic Design for the Gyroscopic Wheel A prototype is developed by [2] and [3] for a single wheeled autonomous vehicle capable of righting itself from any position, spinning about its own axis, moving forward and backward, and avoiding obstacles in its path. The platform gained feedback from the environment using a tilt sensor and electronic compass for both balancing and heading. It also included speed detection and object avoidance by using sonar sensor and shaft encoder on the main drive motor. They demonstrated experimentally that the wheel can automatically be controlled by using the learned human control input. The cost of such experimental setups might represent a burden for the investigators in this area. This work is targeting simulation tools and techniques that might result in lowering such price tag by using virtual prototyping and real time simulation in controlling such system under different manoeuvring tasks using intelligent control scheme. Visual Basic is utilized as a medium for integrating all of these components. A neural network can approximate the response, but is not capable of interpreting the results in terms of natural language. Therefore, using the neural networks and fuzzy logic in the controller design via neurofuzzy would provide both learning and response readability.
Important Links:
Go Back
