COMP4411: Experimental Robotics

• Course introduction
• COMP4411 wiki: http://www.cse.unsw.edu.au/~cs4411/wiki/
• Lecture Slides
• Project Slides
• Assignment 1 files

Completed Projects

• Puzzle Arm: A robot arm/camera combination that can solve jigsaw puzzles.
• Volksbot Camera Laser: Navigation and obstacle avoidance using Omni-Directional Camera. Turns the omnidirectional camera on the Volksbot into a laser-like device.
• Urban Grand Challenge A: A mini version of the DARPA Urban Grand Challenge.
• Urban Grand Challenge B: A mini version of the DARPA Urban Grand Challenge.

2006 Session 2

• Person following using the Pioneer
• Face recognition

2006 Session 1

• Using ZigBee for Localisation/Navigation: Robot localisation using trilateration techniques with a low-power wireless transmitter.
• The Barbie Segway: A self-balancing two-wheeled robot.
• 3D-SnakeBot: SnakeBot capable of turning, with wireless communication and vision.
• Co-operative SLAM: SLAM with multiple robots
• Volksbot Slalom: Control of a Volksbot through a slalom course, directed by using panospheric camera images.

2005 Session 2

• Autonomous Hovercraft: An RC hovercraft with mounted web cams that autonomously steered down corridors using different vision algorithms.
• Gesture Control of Pioneer: A pioneer whose behaviour could be controlled by hand gestures.
• Hexapod: A six legged walking robot.
• One Grippered Bandit: A pioneer that searched for, and retrieved, coloured cans.
• SLAM Comparison: A comparison of two SLAM techniques, DPSLAM and PMAP using Player and the pioneers.
• SnakeBot: An eleven-segment snake robot that learned to crawl on a treadmill via the application of genetic algorithms.
• TuxRacer: An RC car fitted with a Linksys NSLU2 board.

2005 Session 1

• Co-operative Localisation: Two pioneers, one stationary the other moving, used visual tracking to move to a precise location.
• Gesture Control of Robot Arm: Hand gestures were used to control a robot arm.
• Micro Camera: A custom-built camera on a pan/tilt unit that could track an object, similar to CMUCam.
• Obstacle Avoidance: A pioneer navigated the level 3 maze at near-maximum speed.
• Topological Navigation: A pioneer navigated the level 3 maze using arbitrary signs as location markers.

2004 Session 2

• Optical flow obstacle avoidance.
• Localisation in front of goal.
• Locating landmarks using SLAM.
• Controlling SCORBOT using gestures.
• SnakeBot: The first SnakeBot project involved configuring the servos, writing control software and teaching it to crawl via reinforcement learning.

2004 Session 1

• Blimp tracking: A blimp was visually tracked and controlled to hover above a point.
• Landmark identification.
• Obstacle avoidance with Pioneer.
• Dynamic digital depth?
• Victim location: A pioneer visually followed a series of markers to a "victim's" location.
• DraganFlyer: A four-bladed helicopter, fitted with gyros that was made to hover autonomously.

2003 Session 2

• RC Car: A remote-control car was fitted with IR sensors that would stop the vehicle in case of imminent collision. The car itself was controlled by a tilting interface that could be attached to either the wrist or the head (on a hat).
• Voice control of Pioneer.
• Overhead camera on RoboCup field.
• AIBO localisation without beacons.
• Visual and laser detection and tracking of AIBO.
• Pioneer office boy. M/UL>