(none) hits in May 2012
 


CV
Contact
Media
My Links
Projects
  ConnectFour
  Differential
  Encryption
  IceCream
  LGUI
  LaserScannerURG
  Middleware
  MyDb
  NeuroKar
  NomadScout
  PThread
  ParticleFilter
  PathPlanner
  Pose
  Robosoccer
  RobosoccerSim
  RobotLocalization
  SonyEVI_370
  UVector
  UsarServer, labbook
  UsarSim, labbook
  V4l2 api
  Viz
  iCat Demo
Publications
Site Map

Search Google

NeuroKar

Description:

Teach autonomous vehicles not to collide with the walls by giving the right example. An autonomous vehicle is controlled by a neural network which learns from the behaviour of the red vehicle that you control.

Short Manual

After starting the program click the simulation environment to start the vehicles. Your controls are:

Turn left  hold left cursor  
Turn right  hold right cursor  
Start/Stop user vehicle  press Space  
Start/Stop simulation  click simulation environment  
Adjust vehicle features  click a specific vehicle for its menu  

The menus will give you further options like visualizing the neural network of an autonomous vehicle or visualizing the sensor values of a vehicle. A manual in dutch gives a more detailed description.

Start Now or Download

To run as java applet in your browser click Simulation Applet (Quick Start). Please note that java applets can not write to disk because of security limitations and therefore some features are unavailable.

To download and run as java application click Simulation.jar (download and start with 'java -jar Simulatie.jar'), then all features are available.

Requirements

Java Environment 1.2 or higher.

Thesis

This project was the subject of my graduation at the
Hogeschool van Utrecht, an institute of higher education. My thesis is available only in Dutch. See JaNet source-code documentation for the implementation and visualization of neural networks.