My first robot taught me about gaits. It was a basic move forward/bump-reverse deal with an ultrasonic sensor. It also had websocket control. It was overall a pretty disappointing robot since it didn't have a brain.
this is dusty since it sits on my desk as an ornamental piece. it used to carry a giant 8-cell nimh battery on top
So the next robot I would build, I wanted it to be able to have "a brain" with regard to more accurate probing (better than ultrasonic sensor) and an IMU. I also wanted to try the Teensy microcontrollers. It was a good choice since I ended up using many of the available ports eg. both i2c ports and serial.
I initially built this robot with the blue 9g servos which was a mistake. Those are too weak and prone to burning out. So I have since upgraded to the MG90D and they are so strong... I can't even tell that the robot is on since the servos don't make noise.
It's hard to see but the IMU is sandwiched between the main board and the body of the robot. I have been slowly working on how to use IMUs. I currently use it for NED/robot is upside down and estimating the pan angles for the depth sampling.
There is also a voltage tap so I can tell what voltage the battery is at.
This is a long work in progress project.
I also worked with ThreeJS to plot the depth mesh pattern that the robot detects by panning/tilting. This is just a visual thing but ideally the robot does it internally with math. It also takes a very long time to transmit this data via serial-websocket with my jank polling.
Below you can see the mesh scanning process it is slow
You can also see here how it works by doing the scan and detecting obstacles though it's still crude.
Comments