Chronological Entries - 2024

1/2/24 - Camera calibration

For using the Intel camera for accurate positioning, we must calibrate it to calculate the intrinsic and extrinsic parameters of the sensor. I won't go into much detail here as I've already done this exact process before in previous entries. In sum, we use a regularly spaced checkerboard to find the parameters of the sensor.

This time around we'll do a slightly more sophisticated method of finding the absolute position of the ball given the 3D position of the camera in our simulated worlds using matrix math.

1/4/24 - Utilizing homogenous Coordinates

From the camera calibration we get also get the camera's K-matrix. We can take its inverse and find the 3D position of certain objects given its distance from the sensor (given by the depth sensor). Boiled down, this is simply a change of coordinate system, super cool!

1/8/24 - Learning about PnP through OpenCV

Since the exact position of the camera relative to the table might change as it is not directly attached to the table, I wanted a way to locate the sensor. After some searching, PnP seems to be exactly what I need.

I started by reasoning out a method similar to how earthquakes are triangulated but then quickly found OpenCV has their own method through solving the PnP algorithm. Given 4 points on the frame (which are the corners of the table that the user selects), and their respective points in the virtual world, OpenCV can calculate the sensor's 3D position in the world!

3/10/24 - Fixing the pulley tensioner on J4

Here, I've added holes for a shaft to constrain the pulley tensioner.

I'll print these new parts soon as the center part does take some time.

3/11/24 - Redesigning the bending bracket for J2 and J3

Since travel past 180 degrees isn't necessary, I decided to complete the bracket so the two belts can't flex the bracket. This fixed the bending issue and kept the bevel gear engaged under load.

3/12/24 - Rewiring the ODrive Board and Beginning Config

3/13/24 - Configuring the ODrives

7/22/24 - Learning how to Interface with the ODrive

7/24/24 - Very Big NVIDIA Jetson Problem

7/26/24 - Starting Physical Robot Control

7/27/24 - Trying to use the Encoders (and Misconceptions)

IMG_4927.mov

IMG_4922.mov

7/29/24 - Working Indiviual Joint Control

IMG_4935.MOV

7/25/24 - Incorporating this into the Simulation and Issues

7/31/24 - Reflections So Far and Things to Fix

CAD / Mechanical:

Find ways to route Servo Wire (especially at the elbow)
Fix the worm gear spacing on J1, there's large slop in J1
Split center body to allow for easier assembly and printability
Certain bearing pockets crack (add more material or add more hole slop)
Add more material on front of Center Body because it's weak (where the tube attaches)

Software:

Find another x86_64 computer to run the simulation (needs CUDA cores)
Work on delay from simulation (I think its due to acceleration caps, but not sure)
Find ideal camera placement (I think above the robot arm pointed at the table)

Electrical:

Neaten up the wiring for the board (design cable routes with ziptie holes)
I'd like to make custom wire harnesses and connecters for everything for the Final Rev
Figure out where to wire servo to chosen computer
Fix J1 Wire support at the center body
Swap power supply for non-adjustable at 48V

Shifted Focus to College! Updates will come soon!

Page updated

Report abuse