Build a robot? No way!

I finally decided to build a robot!

And i decided that I would blog/vlog the whole endeavor from the start. In this very first post I will try to summarize the scope of the project. What do I really want to build?
  • I want a fairly large robot that is robust to shock, impact and and that can be used outdoors in all sorts of weather and temperatures (I live in the rainiest city in Norway, so that's a given).
  • I would like the robot to last something like 24 hours on one charge without being dependent on any external sources of power.
  • I want the robot to be able to act on it's own to the largest degree possible, that is, I will strive to make it autonomous.
  • I want it to have legs. I like spiders, but having 8 legs has no obvious benefits while adding cost. 6 legs will provide a similar degree of freedom, and it works for insects so I guess it should work for my robot too. In other words, I am making a hexapod.
There all done!

Seriously though, if you were paying any attention at all, I just described a robot so complex that even NASA would be reluctant to build it. This project will take a lot of time and effort, and will most likely never complete, but hopefully it will provide me with the kind of entertainment and hard-earned knowledge that I am addicted to.

So over to what I'll need to get started.
  • I need a hardware design.
  • I need a software program to drive the robot .
  • I need tools to craft the robot body.
  • I need materials to create the robot body from.
First things first: the hardware design. The plan is to create a computer model of the robot body that is refined to the point where it is ready for prototyping. In the prototyping stage the design will be tested against reality and refined some more until I find it's ready for freezing. An essential part of this is sourcing the parts with the right properties to actually being able to implement the plans. This will be a huge equation of a bazillion variables that I will have to juggle for a long time until it sits right.

Then there is the software. I work as a developer and writing software is a second nature to me (first nature)? But still, the software that goes into a mobile walking autonomous robot is not beginner material in any way. Reliable and robust real-time calculation of multiple DOF limb motion for 6 legs, processing of sensor data from countless rotary encoders, the IMU, and real-time processing of several video sources for stereo vision, and we haven't even started with all the subsystems required for NLP, AGI, SLAM....

Then I will need a set of tools to build this robot, and a place to do it. I have a few tools already, but I will probably need a lot more for this project to get anywhere.

Finally I will need to get hold of the necessary materials for prototyping and building the robot.


Now that you know what we are up against, I'll let you ponder until the next post, in which I will give you some pictures of the current status.

PS: I decided to name the robot "Devol" after the father of industrial robotics, George Devol.


No comments:

Post a Comment