I heard about many companies in this space that have since shut down / become acquihired. Root AI acquired by AppHarvest [1], Abundant Robotics (shut down) [2], Traptic acquired by Bowery [3].
To me, this suggests that the problem is hard. Last I checked, the state of the art in robotic grasping seems similar to the state of many other AI systems before ML hit the scene. It's super-mathematized all to the questionable end of analyzing how a few points points can optimally apply forces to simple convex polytopes.
A similar feeling exists when you look at the state of path planning for robotic arms. There, collisions must be avoided at all costs because we don't have the mathematics for it. So you make this super precise plan that carefully snakes its way around all the little voxels that happen to become occupied in your occupancy grid. To execute these plans we need to manufacture robots with expensive harmonic gearing and sub-millimeter level repeatability. These types of robots would not be economical for outdoor picking tasks.
To make progress, I think there will have to be new ML techniques and new lower cost robotic hardware developed in tandem.
The higher value crops (fruits, veggies) have different and more difficult problems than maize,soybean,wheat,cotton which are grown in huge fields with known parameters (thanks to precision planting, etc.). Having machines till the space between these rows would be an easy proof of concept. Syncing and coordinating multi machines in one field would be step two. There's 200M acres like this and farmers would be happy to pay between 10-100 _per acre_ depending on frequency and task performed.
The problem is the "easy crops" are already harvested by a machine that's GPS guided and nearly infinitely wide; at that point hiring some farmer to "pilot" it is cheap.
True - and an important distinction. Hiring someone for $20/hr to sit in the cab of some multi-million dollar harvesting machine isn't that hard.
I think it's a bit of an area that doesn't really benefit from more efficiency (except on the things that have to be picked by hand) - farmers don't even bother flattening their land to make it easier to plant/harvest, as the machines handle slopes and hills just fine.
I agree it is an area that we have probably already reached peak from labour efficiency stand point. Ofc, machines can get slightly bigger and methods slightly more optimised. But there is good reasons to keep a human in loop and near to fix any immediately fixable problems. Or just to fill fertilizer or empty the load.
To me, this suggests that the problem is hard. Last I checked, the state of the art in robotic grasping seems similar to the state of many other AI systems before ML hit the scene. It's super-mathematized all to the questionable end of analyzing how a few points points can optimally apply forces to simple convex polytopes.
A similar feeling exists when you look at the state of path planning for robotic arms. There, collisions must be avoided at all costs because we don't have the mathematics for it. So you make this super precise plan that carefully snakes its way around all the little voxels that happen to become occupied in your occupancy grid. To execute these plans we need to manufacture robots with expensive harmonic gearing and sub-millimeter level repeatability. These types of robots would not be economical for outdoor picking tasks.
To make progress, I think there will have to be new ML techniques and new lower cost robotic hardware developed in tandem.
[1] https://www.appharvest.com/press_release/appharvest-acquires... [2] https://www.therobotreport.com/abundant-robotics-shuts-down-... [3] https://techcrunch.com/2022/02/16/following-acquisition-by-b...