Digital Engineering Project - Flying Swarm
Everyone who wants to participate in one of the projects needs to attend the first meeting!
The first meeting will be done on April 12th at 15:00 in the SwarmLab (G29-035)
In Summer 2017, we offer two DigiEng projects:
Flying Swarm - ROS Interface
- Group Size: 4-6
- Estimated Duration: 03.04.2017 - 30.09.2017
- Required Knowledge:
- Programming in C++ or Python
- Desired Knowledge:
- Network Communication Systems
- Robotics
- ROS (Robot Operating System)
Description
The autonomous swarm of flying quadcopter FINken developed at the faculty by the Intellligent Systems Group were enhanced with a linux compatible embedded PC. This enables the deployment of ROS on each FINken. However the used microcontroller and its software does not directly support ROS as a communication interface. The Benefit of such an integration is tremendous as ROS provide a wide range of toosl to ease development of complex robotic systems.
Goal
This project aims to combine the management, monitoring and configuration mechanisms of ROS with the swarm behaviour of the FINken. The resulting system shall allow a faster and easier prototyping an development of swarm behaviour for the FINken. Additionally, the used ROS mechanism shall ease the access to experimental results.
Tasks
- Deploy ROS on the used embedded PC Odroid C0
- Design and implement a communication interface between the used microcontroller and the Odroid C0
- Map the FINken data to ROS Structures
- Enable the usage of the common robotics tools of ROS: RViz, RQt and Robot-EKF
- Extensive Testing
Flying Swarm - Swarm Behaviour Evaluation
- Group Size: 4
- Estimated Duration: 03.04.2017 - 30.09.2017
- Required Knowledge:
- Programming in C
- Soldering and mechanical composition of robotic systems
- Desired Knowledge:
- Data Analysis in SciPy, Matlab or Octave
- Programming in OCaml and XML
- Swarm Intelligence
Description
The FINken swarm of autonomous quadcopters currently support only a single swarm behaviour. An evaluation of this swarm behaviour is difficult as the behaviour of the robots and the behaviour of the swarm algorithm cannot be seperated. Multiple swarm behaviours enable an in depth analysis of the different characteristic behaviours created by the robots and the individual swarm algorithms. This is the baseline for further optimization of the swarm algorithm towards deployment in professional setups.
Goal
Implementation and evaluation of multiple swarm behaviours on the existing FINken robots with in-depth analysis of generated movement data, communicated information and behaviour decisions. The aim is to infer the characteristic parameters of the robots as well as the swarm behaviours to enable a divide-and-conquer based optimization of swarm-algorithm and robot independently
Tasks
- Analysis of current swarm behaviour
- Implementation of further swarm behaviours (at least 2)
- Extensive evaluatiion of the swarm behaviours in real experiments
- In-Depth analysis of the generated experimental data
Materials: