Robotics
Things that move.
Mobile robots, manipulators, and anything that goes underwater. The part I enjoy most is the bit where it stops being a simulation and has to survive contact with the real world.
Three axis arm
The same arm from the front page. Each slider drives one joint, and the readouts are the live angles. Leave it alone for a few seconds and it goes back to its own routine.
Running its idle routine. Move a slider to take over.
The real world does not care about your assumptions
Most of what I have learned came from sensors that worked perfectly on a desk and then went into a waste bunker. Dust settles on lenses. Vibration walks a mount out of alignment a fraction at a time. Something parks in front of your view. And the data itself quietly degrades in ways that do not announce themselves.
None of the fixes turned out to be clever. They were simple, once we had actually found every problem. That is the real lesson, and it is why I now test far more than seems reasonable.
What I have built
Sensor boxes that have to live outdoors and keep working. A run of smaller robots at university. Underwater vehicle models and control, at the Alfred Wegener Institute and at DFKI, including a library of dynamic models for AUVs built to drop straight into ROS.
These days the robot I think about most is a crane, which is not the shape most people picture when you say robotics.
Building this out
- Write-ups of real builds, with photos and video
- A 3D robot scene you can orbit and drive
- Inverse kinematics: click a target, watch the arm solve for it
- Sensor simulations, so you can see what lidar and ultrasonic actually give you
- A real car you can drive from this page, with a camera watching it live