SHOW ROBOTICS

Starting with zero PCB experience means a long process of asking for advice, making changes, and checking again. I’ve thought “that should be it” a few times already 😅 Another round on MOSS Core, the custom board for my litter-picking robot. @markfrancisio suggested larger test points, and @solderableinc recommended putting one on every power rail. I’ve added 14 new measurement pads and enlarged the existing ones, making it easier to get a multimeter probe onto the board when testing starts. I also worked through more findings from @copperheadhq: moving two tiny holes out of the solder-paste areas near the ESP32, and shortening a capacitor connection for the audio amplifier. The connectors now have clearer pin-1 markings on both sides, too. All the components stayed in place, so these improvements preserve the assembly work already quoted by @PCBWayOfficial. I’m sending them the updated files. A lot of this probably seems obvious when you’ve designed boards before. For me, it’s something I learn because someone takes a few minutes to look and explain. Thanks for helping me get this first board into better shape!

I've been playing with two cheap sensors I'd like to try on MOSS, my litter-picking robot. I printed a little rig for them, found a camera dolly that fits underneath, and spent the afternoon pushing it around the workshop. I'm testing them separately. MightyCam uses VIO, or Visual-Inertial Odometry. Basically, it combines camera images with motion sensors to work out how far it's moved and turned. It works without GPS, but it can drift. Mine seems to follow turns pretty well. The distances are still off, though, so I wouldn't rely on it to navigate MOSS yet. The other one is ST's tiny LiDAR, with about 2,300 depth zones. So far I've just tried it with my hand. I haven't decided where it belongs on MOSS, but maybe next to the RGB camera on the gripper? It could help judge how far away an object is and get a rough idea of its shape before picking it up. Still just experimenting. These little sensors are affordable enough that I can try a few ideas and see what works.

I don’t mind if MOSS, the litter-picking rover, misses a can now and then. I just don’t want it spending the afternoon picking up leaves. I let YOLO spot things on the ground, then showed Jev 258 close-ups and asked, “Would you pick this up?” It said yes to 86 of the 146 bits of litter, and no to all 73 things I’d have left on the ground. Then I gave it five rounds to find 50 pieces of litter. Each time it passed, I showed it a different photo. It got to 49 and never went for any of the 10 decoys. Next I want to roll past the same wrapper twice and see if a different angle helps MOSS spot it. github.com/metrox-eth/moss/t…

My first job was graphic design, and preparing this PCB for MOSS, the litter-picking rover, feels surprisingly familiar. It reminds me of prepress: once the files reach the printing press, one wrong phone number can ruin an entire print run. That made me take another look before manufacturing. I'd forgotten two connectors for the cooling fans, found some silkscreen mistakes, and realised the edge computer's power supply didn't need such a bulky fuse holder. Fixing it freed up some room for the MOSS logo too 🙂 Working with Codex and KiCad made those changes feel manageable. The old order is now cancelled and the updated files are with @PCBWayOfficial for review. To my beginner's eye, it looks good now. But I'm still learning, and I'd love some experienced eyes on it before the first boards are made. The design files are on GitHub, feedback and reviews are very welcome, link in the comments.

I'm excited to find out what a custom PCB can do for MOSS, the litter-picking rover. I designed this board with Codex, using KiCad. Quite an adventure for a hobbyist! @OpenAI @OpenAIDevs Honestly, I'm not sure I would have taken this on if @PCBWayOfficial hadn't reached out about sponsorship. I spoke with @PCBWaySerene this morning, and it's starting to feel real. The files are submitted for review, and I'm sending her the prototype request details. I'm keeping my expectations grounded. Hardware needed testing and revisions long before AI came along, and I'm prepared for this board to need a few too. I've seen other makers get working PCBs with AI's help, which makes me very curious to try it myself. The aim is less wiring, easier assembly and, eventually, lower costs if enough people want to build MOSS to make a batch worthwhile. A more accessible little platform that can move around, manipulate objects and, hopefully, pick up plenty of litter. Thanks to @PCBWayOfficial for giving me the chance to explore this. Now comes the interesting part: finding out what works on the actual robot.

Some news from today. @PCBWayOfficial reached out to sponsor me, so I did a new prototype: MOSS Core V0.7, a custom PCB that replaces most of the modules inside MOSS. I'll send them this as a proposal. @DiracRobotics delivered the simulation. It looks really good, now we need to calibrate it on the real robot, obviously they can't do that part without a robot in their hands. At this stage we can't send robots to our partners yet, but as soon as the finances allow it, it's a priority. The @ST_World STEVAL-VL53L9 module is finally on its way. It's a tiny 3D lidar with 2.3k zones (54 x 42), 5 cm to 9 m range, for about 80 bucks. Really excited to run some tests with it. A lot of gratitude for everyone helping the project, the partnerships we have are really great. It feels good to see sponsors reaching out spontaneously like this, not only for the hardware, it's also very encouraging for the project.

Every battery I bought for MOSS, the litter-picking rover, seemed to fall short on power and runtime. So I built my own, using twelve EVE 50E 21700 cells, the same ones I use in Vita, my robot streamer. Build guide in the comments.

MOSS the 3D printed open source litter picking robot on the beach for the first time. Tracks are well sized for sand: dry sand low dunes, damp sand and the waterline, no problem. The big dry dunes near the entrance are another story, the ground clearance is low and it got stuck twice. A quarter of the way down they get smaller and it passes without trouble. To take the big ones it would need a bigger robot or more ground clearance. It does not drive straight yet, I correct all the time, same as on the roadside video. That one is on me: the two tracks are not tensioned the same. Two tensioners per track, one sets the idler angle and the other the tension, and I only set them by eye after mounting. The encoders agree: on the road both motors turned the same within 1 %, so the drift is between the track and the ground, not in the motors. Battery: 12.1 V at switch-on, 11.8 V at rest back home. The 21700 cells are on their way. The rover was designed in Fusion with Codex in just a couple of weeks. @OpenAI @OpenAIDevs

First roadside drive for MOSS, the 3D printed open source litter picking robot. Tarmac, slopes, dry grass, fresh grass, the tracks go through all of it like a small bulldozer, nothing jams. Driven from the pad, straight from the battery switch, no laptop in the loop. Numbers from the onboard log: over 8 min of driving, over 70 m on each track by the encoders. Pack at 11.68 V at the start, 11.34 V at the stop, roughly 9% of the pack by the resting voltage afterwards. And only about half of the pack is usable before the tracks start stalling in pivots. The motors and the hardware in general held very well. Zero stall guard trips, zero dead pivots. The battery is the weak part, a $20 3S4P pack, 12 cells, sold as 40 Ah and clearly not. Charger and pack get replaced next. Rough estimate with proper high capacity cells: about 50 min and 430 m per charge.

MOSS hardware update Among the 80 revisions, the most important are: - The hull now prints flat on its base, with room for a bigger battery. - I added a speaker and an amplifier (I found out the arm’s camera module already has a microphone). - The hull also comes in two halves for 256 mm beds, for people with a smaller printer. - New 90:1 motors with a 10 A driver, it climbs the ramp now. - Deeper guides on the tracks so they stay on, and one small key at the nose that takes either camera. Last but not least: the overall cost to build is coming down compared to the Jetson version. I’ve received the Raspberry Pi from @OpenAIDevs and its adapter is printing alongside the new hull.

Unboxing a package from @OpenAI 📦 This will go straight into the most affordable version of MOSS the open source litter-picking robot. Watch the video: there are some goodies! Thanks @OpenAIDevs, I am excited to build with Codex!