The Trash Man Cometh
December 18, 2010



In an effort to combat litter and preserve our environment, we have tried to model how pedestrians distribute litter along a path. By creating this model we can run an optimization search on the results to find optimal trash barrel placement.


Concepts Demonstrated

  • A Markov Decision Process, value iteration is used to model people walking throughout a given space littering.
  • A* search like algorithm with backtracking is used as an optimization tool once a litter disbursement model has been created.


Our goal with this project is to help building managers and maintenance staff find optimal placements for barrels. We have seen that the barrels on Olsen 300 keep getting moved because someone thinks they would be better placed somewhere else in the hall. Instead of going through an exhaustive trial and error trash barrel movement process we hope our model shines some light as to where barrels should really be placed such that they provide the most effectiveness.

Technology Used Block Diagram

Evaluation of Results

So far our project has found an efficient way to model trash in a given area as well as provided a very nice optimization search algorithm. For example, when we ran our program on Olsen 300 it consistently told us to place trash barrels in the labs rather than in the hallway. Logically, this makes sense since the labs have more space. We hope that with some more tweaking of our program we can produce some better models so that maybe we can finally solve the litter problem in Olsen 300 or even the University Ave. Bus Stop!

Additional Remarks

It has been very interesting trying to model human behavior. It is truly fascinating how we will drop trash if we see trash near it or how we are sometimes too lazy to walk to the nearest barrel to dispose of our trash. I hope with time we can apply this model to recycling and take our data and apply it towards our sustainability and survival.