Swarm Robotics Studies @ KOVAN Research Lab

1
Swarm Robotics Studies _at_KOVAN Research Lab

• Erol Sahin, Onur Soysal, Ali Emre Turgut, Fatih
  Gokce, Levent Bayindir, Hande Celikkanat

KOVAN Research Lab. Dept. of Computer
Engineering Middle East Technical
University Ankara, Turkey http//kovan.ceng.metu.e
du.tr/
This work was partially funded by the Project
104E066 by TUBITAK (Turkish Scientific and
Technical Council).
2
Topics

• KOBOT A mobile robot designed specifically for
  swarm robotics research
• A Macroscopic Model for Probabilistic Aggregation
  in Swarm Robotic System Bu sunum calistayda
  yapilmadi. Dileyenler sunuma http//swarm-robotic
  s.org/SAB06/ adresinden erisebilirler.

3
KOBOT A mobile robot designed specifically for
swarm robotics research

• Ali Emre Turgut, Fatih Gokce, Hande Celikkanat,
  Levent Bayindir, Erol Sahin

KOVAN Research Lab. Dept. of Computer
Engineering Middle East Technical
University Ankara, Turkey http//kovan.ceng.metu.e
du.tr/
This work was partially funded by the Project
104E066 by TUBITAK (Turkish Scientific and
Technical Council).
4
Wishlist for Swarm Robotic Systems

• Sensing and Signaling
• Interference among robots
• Interference from environmental factors
• Kin-detection
• Stigmergic sensing and signaling
• Genereric sensing

5
Wishlist for Swarm Robotic Systems

• Communication
• Wireless communication between the robots and a
  console
• Wireless communication among robots
• Wireless programming
• Parallel programming

6
Wishlist for Swarm Robotic Systems

• Physical interaction
• Power
• Size
• Cost
• Simulation

7
The Kobot Robotic System

• Robot platform designed for swarm robotics
  research
• Light, small, extendable
• Power efficient (10hr of operation with one cell
  Li-Poly battery)
• Cheap
• Diameter 120mm (CD-sized)
• Weight 350gr with batteries

8
The Kobot Robotic System
9
Short-range Sensing System

• Novel low-power IR-based sensing system
• Kin-detection, obstacle detection
• 7 discrete distance levels in sensing
• Low-interference
• 8 sensors mounted 45o apart in a circular PCB
• PIC12F683 on each sensor
• PIC16F877A for coordination of sensors

10
Block Diagram of a sensor

• Emission power of IR LED adjusted by PWM and
  low-pass filter
• IR signal modulated at 38kHz to minimize
  interference
• Op-amp used to drive IR LED
• LED voltage measured to ensure the settling of
  desired V
• Standard 38kHz receiver to detected incoming IR
  irradiation

11
Operation of a Sensor

• Sensor operates on three states
• Kin-detection Sensor scans environment to detect
  existence of a modulated IR signal. Detected
  signal indicates presence of a kin-robot
• Proximity-sensing An obstacle detection is
  performed for different levels (distances) of led
  power. Minimum level gives the distance.
  Measurement performed when no IR irradiation
  detected
• Data transmission Result sent to main sensor
  controller via serial port

12
Sensor Characteristics
Setup for the experiments
Experiment-1 Range test of an obstacle
Experiment-2 Robot detection test
13
Sensor Characteristics
Average distance values obtained from the range
test of an obstacle
Average distance values obtained from robot
detection test

• Variance is 10 mm
• Very low robot-to-robot interference!

14
Sensor Characteristics
Success rate in the detection of a robot
15
Communication System

• Communication system facilitates
• Wireless communication between the robots and a
  console
• Wireless communication between robots
• Wireless programming
• Wireless parallel programming

16
Communication Hardware

• XBee wireless modem as communication hardware
• IEEE802.15.4/ZigBee protocol support
• Point-to-point, point-to-multipoint and
  peer-to-peer network support
• Serial interface up to 115.2kbaud
• RF throughput at 250kbaud

17
Wireless Parallel Programming

• Host program running on console
• Bootloader running on PIC16F877A
• Wireless parallel programming
• Programming at 57.6kbaud

18
(No Transcript)
19

• Thanks for your attention.
• http//kovan.ceng.metu.edu.tr/
• http//swarm-robotics.org
• We would like to thank Dr. Bugra Koku for his
  help in this work.

This work was partially funded by the Project
104E066 by TUBITAK (Turkish Scientific and
Technical Council).