MAS applications

I found this journal Practical Applications of Agents and Multi-Agent Systems and their list of publications and conference proceedings here.  They look like a very neat conference.  I took a look at their most recent proceedings and found to my delight that they have some cool applications that I’m interested in and make me think that some of my ideas for applications for the Bounty system aren’t so bad.

First, I think that my idea for using the bounty system for reducing traffic problem in an earlier post might be a good one (here and maybe here but it seems like maybe they could be developed a bit more).  Or at least that the bounty model should be applied.  I found the paper, From Goods to Traffic: First Steps Toward an Auction-Based Traffic Signal Controller, that was published this year in paamas which shows that there is some interest in using these sorts of methods.

The other idea I had a while back was to use the bounty system for parking (here)!  They have a paper, Evaluating the Social Benefit of a Negotiation–Based Parking Allocation, which looks into the problem of parking in cities and methods for doing this and how it can help both the city and the drivers.  So, again there is a body of work (they cite) that is interested in this problem.

Maybe I should put together some of my ideas for applying bounties to these problems and try and publish them there?  Also, I’m excited to find out what other problems that paamas has published.  Glad I found this and can’t wait for the semester to begin :).

 

Windows and light

So, I was just reading a Fast CoDesign article about a new product that is essentially a moving mirror that tracks the sun to redirect the light to dark places in a room.  This seems brilliant (haha pun intended).  This made me think that instead of needing this thing in the room wouldn’t it be better if you could coat your window with a substrate that had reflective qualities that would better illuminate or darken your room.  This brings up an interesting point.  I’ve seen in movies and tv shows were the windows are able to go from transparent to opaque.  I have not seen any window amplify or focus light on command or at all besides with a magnifying glass…  So, that would be an interesting optics problem.  I do realize that windows already diffuse daylight by having a coating on them.

I was then considering the fact that the placement of windows for optimal sunlight distribution and heating and cooling effects throughout the house is a challenging optimization and architectural problem.  Basically the idea is that neither solution above would be necessary if the placement and style of window and house were created with that problem in mind.  The style and location of a house/building has a profound impact on the quality and quantity of light that enters the house and is useful.

I must remember this when looking for a house and also possibly if I ever build a house.

Extreme Sports, Baby Cribs and Pipes

In the future we will have new extreme sports.  We are already sort of seeing it with that person that did a jump from close to outer space.  We will start to have ziplines that you can ride from the moon to the earth.  We will surf/swim on the lakes of lava on the sun!  We will skate on astroids.  We could build ice-rinks or glass/transparent swimming pools that float through space!  I am sure there are things that we haven’t even invented yet or dreamt up that will cause us to create new and insanely awesome sports.

On a much more mundane note, we should have a computational geometry person and a mechanical engineer redesign portable baby cribs.  Their current design is terribly difficult to setup and take down.  It should be as easy as pressing a single button or even no buttons!  There shouldn’t be these set of buttons that need to be pressed in order for the thing to be taken down.  It is terribly difficult for all of the buttons to be pressed correctly for everything to work.

What about pipes that don’t clog and if they do there would be this thing inside the pipe that could chop and push the stuff through.  No chemicals needed.  Or if you want chemicals it would be cool if the pressure on the sides of the pipe due to the clog would release a chemical that would dissolve the clog.  The chemicals would be based on some algae that could live on the pipe.  So, you wouldn’t need to dump drano or stuff down the pipe.  Also, this make me think that the pipes in your house could be modeled after the human digestive system.

Airbnb

Not that I would ever do this, but it seems like airbnb would be a good way to scope out places to rob.  You get a nice view of the interior and you get a nice timeline of when the owners are gone.

Adaptive Data Analysis

Google Research recently wrote a piece The reusable holdout: Preserving validity in adaptive data analysis  (so I’m not go to write much).  It details the problem with statistics generated when the machine learning methods are adapted to the data through data analysis and repeated trials on the same hold out data.  This is a problem that is easily recognizable in machine learning competition leader boards like those on Kaggle (good article on that).  The solution they gave was to use their method detailed here and here (going to be published in Science) that allows us to reuse the hold out data set many times without loosing its validity.  So, that is awesome!  Hopefully the Kaggle competitions and data science research will be greatly improved by having more meaningful feedback.

Directed Reading

I’m planning my directed reading class this coming semester.  So, basically I have/get to come up with an entire semester’s worth of material.  I might be able to make a class out of it by the time I’m done 🙂 haha.  My subjects are focusing on the areas I want to explore with the bounty hunting task allocation model.

  1. Petri-net models and applications to MAS and task allocation.  Basically I’m thinking that petri-nets can be used to create task allocation problems abstractly.  By doing so I can better gauge the usefulness of different task allocation methods.
  2. Contest Theory and tournament theory.  They are two areas which seem like they have crucial information for helping the bounty hunting method mature.  Explore how aspects such as effort can be incorporated into the task allocation problem to make it more realistic.
  3. Coalitional Game theory and the El Farol Bar problem.  The game theory aspect will help me create better learning algorithms for the bounty hunters and the El Farol problem is very similar to the bounty hunting problem.  Also, look into collectives, a complex systems term that describes a system of selfish agents that have a well defined system level performance measure.
  4. Specific problems in Multiagent Task Allocation.  Namely cooperative, resource dependent, heterogeneous tasks and tasks that have sub-tasks.
  5. Automated Task decomposition for MAS.  This is mainly so that I can explore the idea that behaviors (and possibly tasks) may be automatically reduced to states with single feature transitions hierarchically using only state action pair sequences.  Doing so will allow complex behaviors to be learned without so much work required by the human to decompose the problem.  I’d also like to apply it to tasks.
  6. Learning coalition formation, or automatic organization formation for MAS mainly with coordination in mind.  Main applications focusing on large scale systems like autonomous vehicles, cloud robotics and crisis management.
  7. Large scale MAL/MAS and applications.  autonomous vehicles, cloud robotics and crisis management
  8. Current topics in reinforcement learning and its use in MAS.

I’m working on fleshing out the details before the semester starts so that I might already have the list of papers I want to read.

Paper citations

I wonder why we only count the number of citations our papers get.  We should also count the citations of all of the papers that cite you, since they were somehow related.  Maybe as the degree of separation increases the weight of each citation in that layer changes.  Would be interesting.  Also, maybe included in the weight calculation is the distance from the related paper in terms of subject area (like astro physics and psychology are light years apart in similarity but if your paper was influential enough it might be cited by someone in psychology and that would imply importance…).

Petri net ideas and Bounty Hunting

I had a rather random idea.  Music seems like it could be modeled by a petri net.  A petri net can model distributed systems, such as the instruments in a band, mathematically.  I googled it and it seems like that I’m not the first to think of this.  I think one of the first papers to model music with a petri net was “Music and Causality” in 1983.  So, seven years after petri nets were formalized they were applied to music.

Could model the system of how the agents are operating in the environment as a petri net.  Going to different tasks and using different resources at a top level to measure system efficiency.  This might be a useful way to look at comparing coordination mechanisms especially for bounty hunting.  Something similar to this idea was discussed on this site.  What if we considered the agents as resources as well…

I might also be able to use them as a deductive device to realize where some inefficiency due to mis-coordination or deadlock are in the system and can develop ideas on how to improve/fix them…  That would be nice to have.  It would also allow me to create custom benchmarks that show that my system is able to solve such problems.  That would be awesome!  This seems like a good idea.  Contract nets did this to prove particular aspects of there system in this paper.

Now I think that the people in Milan Italy are really interested in Petri nets.  I just need to find out how I can go to Italy :).

Gamification of Gate Creation :)

David just connected a wii-remote to one of the darwins and is using the gyro info from the remote to adjust the hip pitch and roll to keep the robot from falling on the astroturf!  Basically doing LFD.  The plan is to create a function that will appropriately set the hip pitch/roll based off of the robot’s gyro data and possibly the foot sensor data using the training data that David gives based off of the wii remote.  Totally awesome!! Hope it works :).