The Advanced Artificial Intelligence Master of the Spanish Open University (UNED), includes a final project of 30 ECTS credits. Concerning to development, SIMBA ( Selective Inspection method of Maps Based in Auctions) is the main part of this final project, the ePuck library and the Virtual Sensor are the other parts. You will find the scientific production at the end of this page.
Robots coordination depends of a reliable communication infrastructure, but there are places (like sewerage, abandoned mines or the underground transport) where this infrastructure cannot ensure a full communication coverage.
Here is where SIMBA plays. When robots need “talk” with others to achieve a determined task but the communication cannot be ensured, we need to coordinate the robots in a way that the final “path” followed to complete the task could be as good as possible.
Short methodology description
Imagine your friends are going to prepare you a cake, a special cake with the best ingredients they will find. For this task, your friends agree to find these ingredients along the city, and they will use the underground transport, but:
- They forget agree about the ingredients will buy each one
- The underground does not offer communication coverage into the tunnels, only in the stations
So, it is very possible that some ingredients will be bought twice, others could be forgotten. What to do, then?, With SIMBA I propose the following:
- When a friend arrives at a station, he/she checks the communication software (imagine it is something like WhatsApp)
- If there is not message informing that the ingredient of this station has been bought, he/she buys it and sends a message to the others
- If the ingredient has been bought, he/she checks what ingredients are missing and decides to buy the nearest
- What happends if two friend meet in the same station or in the chat room?, they agree the ingredients to buy
This is a very superfluous explanation, but it should give you a clue about the main workflow. I don’t want to bore you with technical details, but if you interested in the architecture do not hesitate to check the papers
Some audiovisual material
Here a video of how SIMBA works into a simulator (built by myself for the experiments). The robots task is to cover all map, visiting all unexplored tunnels (showed in black). In the above example, in all underground station your friends will be able to communicate, but in this experiment, into the ‘squares’, the communication probability is 0.5
This project was long and hard. Long because was needed implement SIMBA both in simulator as real environment, and hard because working with real robots always means you will have to solve a lot of problems you did not expect.
Finally, the project was qualified with honours (wow!) and the Spanish Open University offered me to record a radio program explaining my work and my experience coursing the Master. You can find this programme in the following link (Audio in Spanish)
Also, the work explained into the article published in Robotics and Autonomous Systems is explained in a interview for the Scientific Magazine of the Spanish Open University
Entrevista para la revista de divulgación de la UNED: La unión robótica hace la fuerza (Español)
- Auction based method for graphic-like maps inspection by multi-robot system in simulated and real environments, Robotics and Autonomous Systems, ISBN: 0921-8890. Manuel Martín-Ortiz, Javier de Lope, and Félix de la Paz. Download it from ScienceDirect
- Hardware and software infrastructure to provide to small-medium sized robots with advanced sensors, Workshop of Physical Agents (WAF), 2012 XIII World Congress, ISBN: 978-84-940469-0-2. Manuel Martín-Ortiz, Javier de Lope, and Félix de la Paz. Download
- Selective Method Based on Auctions for Map Inspection by Robotic Teams, Foundations on Natural and Artificial Computation. Lecture Notes in Computer Science. Springer Berlin / Heidelberg, ISBN: 978-3-642-21343-4. Manuel Martín-Ortiz, Juan Pereda, Javier de Lope, and Félix de la Paz. Download it from Springer
- Inspection method based on multi-agent auction for graph-like maps, Nature and Biologically Inspired Computing (NaBIC), 2011 Third World Congress on, IEEE. Manuel Martín-Ortiz, Juan Pereda, Javier de Lope, and Félix de la Paz. Download it from IEEE