Project by George Lejnine

The C.A.R. (Collision Avoidance Robot) uses Parallax’s Basic Stamp 2 microcontroller to power a robot, through any course, and given enough time for the user to drive it using a PC interface, the robot will learn and adapt. The idea behind this is fairly simple, create a simple algorithm that will guarantee no crashes, and will adapt over time, while deciding whether the user is right or wrong, and overrides it if wrong otherwise, using its adaptive properties it makes a route for the route.

The baby stage of the robot (the first time it is powered on) has no data on how the conditions are, but it has the ability to avoid any obstacle. After driving it into some near crash encounters, it learns to adapt to the drivers style by deciding where to go in case there will be a crash. This algorithm, although not autonomous, is very versatile and can be used in many large scale items. These include: motorcycles, forklifts, cars, etc. What makes this impressive is that even with a few infrared sensors, it will avoid any obstacle it has to face. In addition for the robot being able to avoid obstacles, it can warn the user both through sound and through the GUI in the application:


Contact: geolej@comcast.net

Downloads:
Collision Avoidance Source Code (.zip)
Collision Avoidance Schematic (.jpg)
YouTube Video (off-site)