A forest robot now picks which trees to thin, with humans watching
A 4.5-ton robotic harvester can find its way through dense forest and thin trees under human supervision.
It's called SAHA. It navigates dense forests and performs selective tree thinning under what the researchers call supervised autonomy.
The machine combines learning-based perception, planning, and control. In field trials, it achieved kilometer-scale navigation and tree grasping at speeds comparable to human operators.
Researchers at ETH Zürich's Robotic Systems Lab built it, working with Universität Bonn and PreFor Oy. The authors include Fang Nan and Marco Hutter, and the work was supported by the EU project Digiforest.
The team describes the results as the first supervised-autonomous thinning operations in challenging forest environments, and says the trials were real-world field trials.
The channel's description gives no figures for trees per hour, trial length, or location. The full paper on arXiv and IEEE Xplore covers the system and results in detail.
Would you trust a robot to decide which trees come out of a forest?
Archive video: NASA; U.S. Army 105MPAD by Spc. Rayonne Bissant, Public domain (Wikimedia Commons); U.S. Army video by Sgt. C Jay Spence, Public domain (Wikimedia Commons)
Video: ETH Zurich Robotic Systems Lab on YouTube