I did a quick assembly so I could test the drive system. Also waiting for high-resolution rotary position sensor for better absolute rotor position. Modifications on the VECS´s are needed. (Happy with the very silent drive)
Friday fun!
Testing the AS5048A encoder with 16384 positions per revolution and with the temporary drive system. Motor brackets are 3D printed and not perfect, therefore some annoying noises! Running VESC FOC Mode @10% Duty.
With the rotary position sensor I get a very smooth start and better torque. Before the start from standstill could be a little choppy and the sensor took care of this issue.
Next week, CNC milling of motor brackets begins.
Here are the final brackets and the plan for gearing
a quick render on the drive system
In the meantime I keep the printer busy with printing pulleys.
Sizes printed today is XL 10T, 50T and a 55T. Took a few test prints before I could find the sweet spot for the settings for a maximum belt grip.
New AS5048A encoder mount printed. HE-14 header fits great and I installed pull relief for the cables on the back.
Saturday’s fun. Printed mounting brackets for the front wheels (Please note that this is only test wheels before I make an effort the design some new ones with bearings etc.
The brackets came out great. Following the contours of the body and blended in quite good. This parts was also printed with PETG.
Testing the 3D printed front nose. Looks quite good. Minor adjustments needs to be done.
Printed with PrimaSelect PETG @240C, 85C bed, 3shells and 15% infill (Very solid outcome).
Countersink need to be adjusted to cut the body.
The Integrated ZR300 camera need some cover on the back.
Check if one can use a 90 degree USB instead of a normal USB head.
The robot got eyes! Before mowing along with the design of the nose I needed to confirm the field of view. Running Intel® RealSense™ ZR300 over ROS and a depth cloud in RVIZ and a image source.
[font=Tahoma, Arial, Helvetica, sans-serif][highlight=#f1f1f1]So yesterday, CNC milling of motor brackets was started. [/highlight][/font]
[font=Tahoma, Arial, Helvetica, sans-serif][highlight=#f1f1f1]The inside contour was finished. The outside contour is left. [/highlight][/font]
Update to the build:
First big body cover printed. The part came out great. Running PrimaSelect PETG Solidwhite at 240C. This print was printed with 0.6mm nozzle instead of the original 0.4mm. layers looks nice and even! This part was printed at 2 shells at 10% infill and print time landed on around 15h.
Thanks CBenson. I start wondering a bit when this forum is so very quiet… Maybe this project is not for everyone.
Update:
olala, Today all parts for the drive system is completed.
Some parts will be welded together and then the assembly will begin next week!
I have to admit that I am a little excited
Test assembly of the rear wheel arch’s. Looking quite cool. What do you think? Part 2 is a bit warped but I think it’s fixable when doing the finishing
Some very basic posts which don’t involve much design or construction sometimes get many views, whereas others which are far more involved like this one sometimes don’t get many.
The content of your post is excellent, showing step by step.
[font=Arial][size=2]This weekend I have also started exploring the ROS world, sensor simulation, mapping and localization.
-3D world, robot and sensor simulation (360 degree LiDAR, depth camera)
-2D mapping with gmapping
-2D localization with amcl
-3D mapping with octomap
-3D localization with humanoid_localization[/size]
[size=2]-3D mapping with RTAB-Map and [/size][size=2]octomap[/size]
[size=2]-Flat world and non flat world[/size][/font]
[font=Arial][/font]
A successful first tests. I will have to spend some time digging down in each package and explore there possibilities. One problem for localization with an appearance based localization (example RTAB) is that small light changes dramatically effect the number of feature points.So in other words, because it’s using the color 2D information of the pixels it’s important the location looks always the same.
So when mapping outdoors I would love to hear what is recommended or what other people have find working good? As mention earlier I am running the ZR300 and sweep LiDAR. [font=Arial][/font]
[size=2]3D localization with humanoid_localization (flat-world)[/size]
[font=system-ui, -apple-system, BlinkMacSystemFont, .SFNSText-Regular, sans-serif][/font]
[size=2]3D mapping with RTAB-Map and [/size][size=2]octomap (non-flat world) btw: the turtlebot is not really good driving outdoors [/size]
So the last days I have been busy. All parts for the drive system was completed and tonight I finished the first test assembly of the drive system. Its a lot of parts for this part of the project and a lot of things need to be correct in order for a nice run. It took a bit of time to get it right and it was a lot of tinkering.
Forgot to mention: Quick example of virtual fences (green outline) using only a 360 LiDAR and AMCL for localization. Virtual fences can be created around objects, trees or whatever. This is a first test setting up virtual fences in RVIZ, later on the fences can be setup when driving the robot around objects.