Marso Hack Berlin 2026 – Robot Parcel Sorting Challenge
Published:
- Developing an imitation-learning solution for a simulated robot parcel-sorting challenge using ManiSkill3, SAPIEN, and a Franka Panda manipulator.
- Parsing and replaying expert demonstration trajectories stored in HDF5/JSON formats across easy, medium, and hard task settings; extracting observations, actions, environment states, and success labels for training and evaluation.
- Establishing behavior-cloning baselines from state and RGB observations, and evaluating advanced visuomotor policies such as Diffusion Policy or ACT to improve generalization across held-out test episodes.
- Building a reproducible training, rollout, and submission pipeline; measuring policy performance by task success rate and preparing competition submissions on Kaggle.