IM 810: A Liter of Computronium

Beep boop - this is a robot. A new show has been posted to TWiT…

What are your thoughts about today’s show? We’d love to hear from you!

Paris hit the nail on the head again, but was completely ignored by Leo and Jeff, when it comes to robots, why make them bipedal? Why make them look like humans?

I’ve said this before, but at the end of last year X was full of tech bros’ videos showing AI representations of bipedal, humanoid robots working in fields, cutting wheat, picking out weeds and harvesting rice, all with captions about how this will revolutionize farming and save farmers so much time…

As if they have totally missed the last 3000 years of advancements in farming, like horse drawn ploughs and harvester-threshers, let alone tractor pulled ploughs and combine harvesters which already do the work 100s of times more efficiently than a bipedal robot could do it…

I pointed that out to one of the people showing the videos and he came up with other examples, like harvesting asparagus, grapes and a few other crops, which are “still done by hand”, I answered with the home pages of manufacturers of machines and robots (not humanoid) that have been doing that for years, if not decades…

Heck, they are using robotic laser firing rigs that run over the fields burning off the weeds.

Another showed sowing seeds per hand from a humanoid robot… Farmers already have machines that can drive over the field at over 10mph and spread the seeds evenly over 10 rows at a time, you’d need to replace that machine with a dozens of the bipedal robots, and the machines sow the seeds at the correct distance and cover them over with earth automatically.

Or how about spreading fertilizer over the crops? A robot could take on that… Except that there are already machines to do that, and they cover almost 100ft width in one pass.

It is as if they are depserate to make a humanoid robot and find some niche for it to work in, without actually thinking about whether a humanoid form is the most efficient way of accomplishing the tasks. They waste so much time on making them walk, when rolling is much more efficient.

Heck, No. 5 was a much more efficient robot than the humanoid robots we have today.

@Leo mentioned a firefighting robot would be better in humanoid form. Why? A robot with a good caterpillar track and articulation would be much more stable on uneven flooring with limited visibility, it could be fitted with tanks and would be a more solid base for carrying a hose. Likewise, a low, caterpillar driven platform with a hydraulic stretcher would be much more efficient and safer than 2 humanoid robots carrying a stretcher.

The human form is a lousy model for making automated things, because we are an evolutionary compromise, we developed from creatures that moved around on all fours and we learnt to stand up straight, which isn’t natural to our form, but we have adapted, but we are a long way from being perfect, I currently have an infection which affects my equilibrium and I am taking drugs to fight it that don’t help, in some ways make it worse, so I am walking around clumsily, if I still could use all fours efficiently, I could get around much better and would have a lower likelihood of falling over and injuring myself…

Edit: Okay, the conversation came back around to it about an hour later, but only briefly covered it, before going on to talk about automated warehouses. These have been around for decades. I worked with a customer around 2012 who already had a fully mechanised picking system for their warehouse - they sold meat products and the cool house could remain sealed and nobody had to run around in thick coats getting products off of the shelves, it was all bins in a 20M high warehouse, the finished products were brought to bins by a set of robotic lifters and the computer knew where each product was and when products needed to be pulled, the lifters went around and pulled individual items out of the bins and placed them in a cage, which went to an airlock, once the picking process was complete.

The first such robot I remember was a silo with tapes in it at the end of the 80s, the mainframe needed the tapes (geophysical plot results) and would instruct the robotic arms to pull the correct tape out of the silo and place it in one of the drives, when it was finished, the robot replaced the tape in a free slot in the silo - they didn’t have specific slots, as they were constantly changing, but the computer knew where they were and the head of the robotic arm had a scanner on it that double checked the correct tape was being retrieved.

1 Like