Multi-Agent Neural Rearrangement Planning of Objects in Cluttered Environments

Vivek Gupta, Praphpreet Dhir, Jeegn Dani, and Ahmed H. Qureshi

Purdue University, West Lafayette, IN 47907

Multi-Agent Neural Rearrangement (MANER) framework addresses the challenge of multi-agent object rearrangement planning in cluttered environments. It analyzes the current and target state of the environment, utilizing visual information to determine rearrangement actions, enabling efficient scene arrangement while considering the complexities introduced by multiple agents operating in the same environment.

[arXiv] | [Github Coming Soon] 

Real World and Simulation Experiments

4obj-real.mp4
6obj-real.mp4
exp1.mp4
exp2.mp4
exp3.mp4
exp4.mp4
exp5.mp4

Questions?

Contact us to get more information on the project