Fuzzy Control for Two-wheeled Self-balancing Robot

Implement Sugeno Fuzzy Rules for Adaptive Controller on a Two-wheels Self-balancing Robot (TWSR)
Full report: link

Identifying Linear Model

Simulation

Set the input signal $u = 0.5 * Step(t-5)$,

Approximating the model

Approximate the model to $G(s) = \frac{k}{s(Ts+1)}$ with $k=9.6250; T=0.388$ Based on the linear model, design an optimal PID controller, we get: $K_p=-6.26684; K_I=-9.07996; K_D=-0.79364$

Conduct the simulation with the PID controller, and inject a noisy pulse to test the stability of the system

It results in a very good performance!

Design Fuzzy Rules

Fuzzy rules

Let $x_d, \theta_d, \Psi_d$ be the desired states when we control the vehicle. Let set $\theta_d = 0$, and the other desired states depend on the desired trajectory. We define the Sugeno model as follows

Simulation

The simulation results are excellent For x:

For Theta:

For Psi:

Implement on a real TWSR

The results are comparable to the simulation.

For x:

For Theta:

For Psi: