AESWARM Mert Doğar Hazar İlhan Dünya Değirmenci

The Robots Two DC motors on the sides, one free wheel at the rear Chassis and wheels made of white polyethylene Run on Nokia batteries, rechargeable Wheels partially clad with o-rings for sufficient grip

The Robots Power Consumption Main consumers of power on the robot are dsPic33: 300mA absolute maximum (average around 100mA) DC motors: 250mA measured maximum (normally around mA)

The Environments Three constraints while designing the environments: – Corridor width, must not cause congestion – Lines of sight, must create different conditions of communication – Charger placement, must allow various scenarios As such, we created three different environments for the robots to live in. – Each fit in an area of 110x97cm – Wall height: 13cm

Future work 2 microphones to capture auditory data 2d localization techniques on a microphone array A CMOS camera to capture visual data Image processing and pattern recognition techniques IR sensor Ambient sensor Voltage divider for battery level Microphone inputs CMOS Camera

Currently Robots are: Able to perform proximity analysis from infrared density Working with outputs derived from SOMs Can select source SOM according to ambient

Extra Memory All robots contain a 2gb non-volatile memory Used for logging Contains all soms(i.e. genes) available to a robot Contains starting training vectors  Training vectors will be overwrited when passing a generation

A milestone: Self-Charging  We want our swarm units to automatically detect low power and start search for charging area.  This will be carried out by  following the nearby walls  Listening to communication buffer if anyone who found the charging area is yelling out the location Charging area

Framework so far:  Simulation of wall following and corridor entrance behaviour on computer environment