Some background info

We are a very fancy non-mainstream team, which have no idea how did it manage to get to this semester, but we will keep marching on.

Some future info

Well... We haven't got any plans yet. BUT! We will MAKE IT!

About our project

This project is about a neuro-fuzzy controlled turtlebot, which has been upgraded with super trouper ultra sound sensors.

The idea of our control system is really simple. We send the required positions a.k.a. XY to the bot, after that, the bot calucates the route and tries to get to that required position without getting in trouble e.g. hitting any obstacle, running through a fireplace or breaking any walls.

GIT
play-icon

Promo video

THE PROJECT

Neuro - Fuzzy

Neuro-fuzzy controller has 4 inputs to the neural network

Neural network

Four-layer perceptron. The output layer has a single neuron, which produces the steering angle to control the direction of movement

Activation function and Method

Sigmoid activation function. Error back propagation method

Fuzzy controller

Inputs to the fuzzy controller are the distances of obstacles from three different angles and the value of the initial steering angle

The final outputs from the fuzzy controller are the velocities for left and right wheel

Fuzzy membership functions

Terms such as near, medium and far are used for left, right and front obstacle

Terms such as positive, zero and negative are defined for initial-steering-angle

Terms such as fast, medium and slow, are defined for left and right wheel velocity

Fuzzy control rules

See source code and documentation

The Fellowship of the Ring

LEGOLAS

Prince of the Woodland Realm

GANDALF

Servant of the Secret Fire

FRODO

Bearer of the One Ring

ARAGORN

High King of Gondor and Arnor

GIMLI

Lord of the Glittering Caves