Get the door! An Example of Cooperative Robotics

My final report: Attach:BenCharrowFinal.pdf

Anirudha Mahumdar

I liked the idea of trajectory optimization and thought the idea of a "library of kicks" was nice. It would be especially neat if, as you mentioned, the library could be searched at runtime. This could benefit if you had a real robot in mind, or more extensive simulations.

Avik De

I liked how thorough the project was and your use of differential geometry to plan through the space. This work could benefit from a discussion of how far away you are from optimal when moving along the path.

Brian MacAllister

I liked the complexity of your simulation. It would be nice if you had a better justification for why the theta dimension is important.

Caitlin Powers

This was a really neat project: I would love to have something tune my parameters online.

Fei Miao

It would be nice if there was a way for both robots to move simultaneously, instead of having each one perform a small step and then the other one, etc.

Jason Owens

I liked your video and explanation of what your detector did and why. This work could benefit by handling different scales better. It seems like your approach will not work well when you are far away from a backpack or close up.

Kartik Mohta

I liked your overall approach and the fact that this works for such a broad class of probability distributions. This work could benefit from a concrete application. Where does the probability distribution that you're trying to maximize coverage of come from?

Menglong Zhu

Your project motivation and approach were both excellent. For future work, it would be interesting to see whether or not the PR2 can get at the semantics of what it reads. For example, to read a map.

Phillip Dames

Your problem formulation is neat; I like how you did exploration in terms of information gain. Short-term I think you could benefit the most by running your simulation to completion and by simulating multiple robots.

Stephen McGill

Great demo and implementation! I liked how thoroughly you broke the problem down. Longer term it would be nice if the control was no longer feed-forward and if you had some actual synchronization between the robots.