The spider army project
Aim for the moon!
“Shoot for the moon. Even if you miss, you’ll land among the stars.” — Norman Vincent Peale.
I’ve got a plan to build an army of robotic spiders. By army I mean more robots than I can directly control at once …and not all of them will be spiders, I have some scratchings of praying mantis and even a wasp.
But why?
For starters, robots are cool!
Ever since seeing the early Terminator movies, I’ve been in love with the idea of making robots. Robotics has some very difficult and complex science engineering requirements making it a virtually endless learning direction.
But really, injuries suck
Many year’s ago I designed artificial magnetic muscles that are meant to replace human muscles — mostly to help people who have mobility problems or people like me with an injury that a) has an impact on their every day life and b) can’t be fixed.
The problem is, the designs I have are well beyond my capacity to build; they will work but I don’t have the skill or the machinery to build them, yet.
With muscular-skeletal machinery that is far beyond my (current) capability in mind, I needed a couple of things:
- Knowledge +
- Skill +
- Experience +
- Machinery =
- Cool robotic implants
But how do you get Knowledge, Skill, Experience and the right Machinery to build cool robotic implants?
You spend an immeasurable amount of time and money making endless iterations of robots and constantly upgrading your machinery until you have everything you need.
Work might help
For a long time I have had awesome day jobs but business in general is about results, short and long term predictable results and short-lived experimentation.
With this in mind, if I went to the business and said “oh hey, I’d like to take an unknown amount of time — probably a lot — to build something that’s probably not going to be useful to anyone, ever but it might have something good out of it in 10 years or so” they’d look at me confused and then assume it was a joke.
It’s also quite probable that work paying for my time to make stuff, that work would want to regulate it and own the idea in the end… which is probably fair.
Let’s not get work to help
So, for me to retain ownership and freedom to be furiously busy and unproductive — this is going to have to be a “spare time and money” project.
This is the path
In my time I have realised that the best way to learn something is to make it a project and put some goals on it.
A plan gives you some thing to work on and goals make it possible to achieve some success mojo without finishing it.
The plan
Build 7 robots in ~15 years so I have to finish the last robot in 2030. Each robot will have it’s own success criteria.
Right.. but spiders you say?
The first question people ask; Why spiders?
The short answer; spiders are beautiful and I love the way they move. Out of all the creatures on the surface of the earth, my favourite is the Spider.
The more accepted answer; Spiders have heaps of legs so its easy to make them balance.
The models of the spider army
The best stories have lofty dreams with complex interwoven outcomes that develop over time. I needed to split up the art of learning to be good at something into a predictable direction so I came up with the spider army’s 7 models, each of them with a specific learning directive that would have me not just understand but really know something about that purpose. Only 5 of the robots are actually spiders but the robotic spider army has a nice ring to it.
Versions
For each of the robots there’s going to end up being a bunch of versions, basically because I don’t know what I’m doing and I need to be able to feel some success after an inordinate amount of failing. A version is a solid thing I can work towards and then replace and it still feel like the same robot.
Versions are roughly measured by how much pain is caused in replacing something essentially when I say a version, I mean I’ve replaced at least half of the software or hardware or both.
It’s not an exact science, after-all I’m a madman who builds robotic spiders and justifies it to himself without any difficulty.
All the robots
Robot 1: Steve the spider
Steve is where it all started, the first robotic spider.
At the time of writing this Steve is at version 27 and I have finished him. I was going to build more versions and make him more stable and demonstrable but I have taken his story and components to a point where he’d be a different robot if I changed him.
Goal: To be controlled wirelessly, with the ability to walk forwards, backwards and rotate both clockwise and anticlockwise.
Sideline goal: Holy shit I built a robotic spider!
At version 15-ish, Steve walked convincingly but was dragging his but, he couldn’t lift his own weight off the ground so I added to my personal definition that walking included being able to support his own weight.
At version 20-ish, Steve walked convincingly and was able to support his own weight, I kept going and improving because the finish quality wasn’t quite there.
Started: 2015
Completed: 2017
Last Version: 27
Robot 2: Yorick the world ender (spider)
He’s currently at version 8 and I’m in software mode because I don’t yet have the skill to make his 12kg chassis walk.
Goal: To be able to walk up a flight of stairs, turn around and walk back down again — the only input is that I told him to walk forward (by voice or remote)
Started: 2016
Estimated completion:2023
Robot 3: Bobbie the bot (spider)
Bobbie is currently in production (started June 2019) Bobbie is a basic walker with relatively short legs.
Bobbie’s goal: To be able to use virtualised training and machine learning to infer leg geometries, walking gaits and waypoint planning —where I can transfer the training outcomes to the robot.
This means I’ll use ROS for the runtime system and have a virtual version of the spider that trains itself to do tasks.
Started: 2019
Estimated completion: 2022
Robot 4: Wrath (praying mantis)
Wrath will be a 1650mm tall praying mantis that can follow me around.
Goal: For wrath to be able to follow me to work, avoiding obstacles, getting on public transport and survive the trip without running out of power or killing all humans.
Estimated start: 2022
Estimated completion: 2026
Robot 5: Kirby the spider
Kirby will be a small spider.
Goal: Kirby can implement the physics required to climb up a wall, wall can be porous.
Estimated start: 2023
Estimated completion: 2024
Robot 6: Weirdo (spider)
This one will basically just be a variant of Bobbie with an articulating spine and a monocular 3d camera for perception.
Goal: for me to have fun and make something weird.
Estimated start: 2024
Estimated completion: 2025
Robot 7: Archon (secret)
Archon will be a robot that people are going to just have to wait to find out about.
Estimated start: 2025
Estimated completion: 2030
Fin.
That’s all for now :)
