ATLAS

THE ARM WORKS! Update: after another 5 Hours of work, my robotic arm now works extremely smoothly and precisely. No matter where you put an object that is an acceptable size (fits in the claw), if it is in the cameras FOV, the arm can pick it up. It does this using inverse kinematics and trigonometry to calculate the position that each servo needs to be at to grab the object. Below is a video of the robotic arm picking up a light roll of double sided tape (and if the video doesn’t work then a photo):

Trying to get the robotic arm and object detection to work, so far I have tried to implement a coordinate system so that my camera knows how far away an object is and what the size of it is so that the robotic arm can pick it up precisely. I made a website where you can calibrate the robot and also control the arm manually, the calibration gives you the size of and object and the x y coordinates in mm. it currently uses OpenCV for object recognition.

Update: As required for funding, I made the schematic of the wiring for ATLAS. I started off with no prior knowledge about KiCad or even what a schematic Is, and in 2 hours, built a full schematic, checked it with electrical rules, connections, and all verifications, and now I have a working wiring diagram that will be helpful when building. Below is a picture of my schematic:

Devlog #2: I made the first version of object recognition with OpenCV on my Raspberry Pi 3B+. The goal of having object recognition on my robot is so that it knows when there is an object that it needs to pick up, and so that it can orient itself to the appropriate position to pick up the object, and pick it up precisely. Below are some pictures of the code that im running and the output:

This is my first design of my autonomous rover, Atlas. In this session, I fully designed V1 of this robot. The chassis of the robot features a 4WD system with Skid-Steering. The drivetrain is powered by 4 seperate 12V motors, geared in a 3:7 ratio using 27 and 63 teeth herringbone gears. The actual base of the robot is made up of 1 flat plate, with a 25x25 grid of 3mm holes, spaced 10mm apart. The goal of this grid system was that the robot should be modular, meaning that you can use the same robot to do multiple things, by designing attachments that you can screw onto the chassis, making it extremely efficient at doing many different tasks. The robot is eventually going to be powered with a Raspberry Pi, controlling 2 arduinos, one for sensor information and another to control motors (drivetrain, servos, etc). The robot is going to use this modular design to mount a 4DoF robotic arm, which will be powered by 4 micro servos, and at it will also use Object Detection powered by the Raspberry-Pi Cam V2. At the end, the robot should be able to autonomously drive around whatever area you put it in, find loose objects on the ground, pick them up, and put them into a fixed location, while avoiding obstacles and ensuring the safety of people/animals around it. This is just a sneak peek at the potential of this robot, and in the future, I hope to make even more out of it. (sorry if this was a little wordy lol)