Reproduce “Sim-to-Real Reinforcement Learning for Path-Safe Target Reaching in Robotic Harvesting of Hydroponic Lettuce” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Al Bashir, Partho Ghose, Yibin Wang, Robert G. Hardin, Joshua Peeples, Genhua Niu, Azlan Zahid
PSA-PPO trains a robot to guide a cutter to hydroponic lettuce stems above NFT channels, accounting for approach direction and clearance. The policy transfers from simulation to a UR10e without policy fine-tuning. It reached 142/150 targets in a controlled benchmark, versus 132/150 for RRTConnect. Greenhouse counts imply 108/144 complete harvests, not the abstract’s 86.1%; clearance checks are not formal collision guarantees.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.