Wheeled Bipedal Robot

LQR based Wheeled Bipedal Robot: Design, Implementation in Simulation

This project implements a Linear Quadratic Regulator (LQR) for a wheeled bipedal robot in simulation. As an optimal state-feedback controller, LQR utilizes a system model and offers improved performance over traditional PID controllers, ensuring better stability and control efficiency.

Current Progress:

• Simulation environment established in pybullet.
• Hardware has been acquired.
• Assembly and transfering simulation to real world is in progress.

esp32 Based Controller

esp32 developmnet module has beeen used here, for all communication, and computational needs.

Simulation in PyBullet Simulator:

Basic Implementation (without white noise)

(with Filtered Gaussian Noise std:0.2)

Plots, of various filters implemented

Overview

• The 3 videos shown above display the implementations of LQR, in both perfect world, and real world scenarios.
• Real world scenario is achieved by adding gaussian noise to all sensor readings.
• 3 filters have been implemented, namely: low pass(exponential average) filter, moving average filter, and median filter. here: LPF + MA filter has been found to give best results.

Technical Details

Hardware

• In development

Software

• Programming Language: Python, embedded C.

Source Code

The complete codebase and documentation are available on GitHub

Acknowledgments

• Design, Software, Hardware, Electronics: me
• Funding: IIT Dhanbad

What’s Next?

A lot of cool stuff.

• Completion of hardware
• Tuning and Testing