Day 4: OrbiBot V1 powered up, then lifted onto a box so the wheels spin free.
Test motors with the wheels in the air: a reversed motor or a wrong encoder sign just spins a wheel instead of driving the robot off the desk. #robotics
Day 3: first electronics, first light.
A Raspberry Pi runs ROS 2; a Yahboom STM32 board drives the four motors and reads their encoders, over serial. Real-time on the microcontroller, thinking on Linux.
And yes, blue underglow. #robotics#ROS2
@atomicmass_ltd Good question. The big one was the welded steel frame. Welding it by hand left the alignment off, the wheels never sat quite straight, and it got heavy. So V2 dropped it for a CAD design with light, laser-cut acrylic plates. Day 16 covers it.
OrbiBot V1: the first robot I built that actually drove. Welded steel frame, four mecanum wheels, ROS 2, LiDAR mapping.
I'm posting the whole build log, one piece a day: V1, then V2, then V3.
Day 1 👇 #robotics#ROS2
@Re11oy Software first for me. I started with Arduino and Raspberry Pi projects, then ROS. Welding came much later, and yeah, it was scary at first. I'd start with ROS 2 on a small kit. The frame can wait.
@Ferbin08 The EKF only takes wheel speeds and rejects sudden jumps. Heading comes from the gyro, accel is off because the rollers shake it, and LiDAR scan matching pulls the drift back. Next I'm measuring each wheel's loaded radius, since front-left counts about 5% more than front-right.
@Ferbin08 Thanks! Same here, sensor bring-up took longer than the frame. Slip is real on V1: driving 0.8 m forward and 0.8 m back left it 12-20 cm off the start every time. So I don't trust wheel position at all.
Day 2: OrbiBot V1 began as a welded stainless-steel frame on four 100 mm mecanum wheels, with an RPLIDAR A1 on top.
Mecanum rollers sit at 45°, so it drives sideways and spins in place without steering. #robotics#mecanum