State space based robotic control design and implementation

• Main Author: Bay, Si Yao
• Other Authors: Vun Chan Hua, Nicholas
• Format: Final Year Project
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/66743
• Institution: Nanyang Technological University

This project presents the development of a microcontroller-based 2-wheel self-balancing robotic system using the state-space modelling technique. The digital control algorithm used is based on Linear Quadratic Regulator (LQR) as it is commonly used for state-space representation based digital control system. The project begins with intensive research and study on system modelling using the state-space representation, and the LQR controller operating principles. The state variable parameters of the 2-wheel self-balancing robot, which is essentially an inverted pendulum robotic system, are then determined. In addition, various components sensing devices and control modules that are to be used in the robotic system are also studied. A 2-wheel robotic system is then built from scratch, and the necessary softwares are then developed, tested and debug iteratively that eventually leads to the fully operational self-balancing robot. Finally, a Smart phone based App is also developed that allows the robot to be wirelessly controlled and manoeuvred. This report begins by discussing the method used to acquire the measurements of the (four) state variables of the robotic system, which are vital for the implementation of the controller. System dynamics are then simulated using Matlab to determine the appropriate feedback gain value required by the controller to balance the robotic system. An overview of the method the controller is designed is next presented, which includes the implementation of the digital control loop described in pseudo-code. Test results and non-optimal performance of the robotic system indicate that there are inaccuracies in the model primarily due to assumptions made for the measurements that are difficult to acquire with the lack of proper measuring equipment. Fine-tuning methods using graphical aids interface and real-time monitoring of state variables were then used to compensate for the inaccuracies in the model despite them being time-consuming and tedious. Several improvements such as pre-compensating for steady-state error and new control signal scaling that take into consideration of the dead-zone level of the motors were made to enhance the performance of the robotic system. The final delivery of the project is a robotic system that is able to balance without much oscillating motion, including the ability to withstand minor nudge from external forces. This demonstrates that the state-space based robotic control design and implementation is successful, despite it being improved by fine-tuning method.