The Sesame Robot Project

Designing the Hardware (and failing a few times)

Designing a robot that walks is tricky. Designing one that is small, cheap, and easy to print is a whole different headache.

I did all the design work in Fusion 360. My first concept actually had three servos per leg. While that offers more freedom of movement, it made the robot look like a spindly spider and blew the budget. I scrapped it and simplified the design to two servos per leg—one for the hip, one for the knee.

Optimizing for Printability

One of my biggest pet peeves with 3D printed projects is wrestling with support material. I spent a lot of time optimizing the Sesame parts (links, body, covers) to print without supports wherever possible.

For example, the femur links have specific geometry changes (squared-off sections vs. rounded ones) solely to ensure they lay flat on the print bed. The internal frame is packed tight; I had to figure out how to cram a battery, a microcontroller, and a nest of wires into a box the size of a bar of soap.

Sesame vs. Sesame Pro

During the design phase, I got in touch with Petoi (a company that makes awesome robot dogs). They sent over some high-end motors, which led to a fork in the project: Sesame Pro.

Because the design is parametric and modular, I was able to modify the legs to accept wheels and high-end actuators without changing the core body. It proves that Sesame isn't just a robot; it's a platform. You can print the version with cat ears, the version that hides the wires, or remix the STEP files to make a six-legged version if you’re feeling chaotic.

Currently, I’m on Version 71 of the design. It walks, it fits together like a puzzle, and it’s finally ready for the public. 

The Electronics & The "Spaghetti" Problem

Let’s talk about wiring.

The working prototype of Sesame is... functional. It uses an ESP32-S2 Mini and a perfboard. But if you look inside, its sort-of a terrifying bird's nest of silicone wire and solder. It worked, but asking other people to replicate that mess does seem cruel

I wanted a cleaner solution. That’s where the Sesame Distro Board comes in.

Going Custom with PCBWay

I designed a custom PCB that acts as a "hat" for an ESP32-DevKitC-32E. It breaks out all 8 servo channels, handles the I2C connection for the OLED screen, and manages power distribution.

A huge shoutout to PCBWay for sponsoring this part of the project. I sent them my Gerber files, and a few days later, I had these beautiful black-and-gold PCBs in hand. They are genuinely some of the nicest boards I've handled. Using a dedicated PCB turns a 4-hour frustration-filled wiring session into a 20-minute soldering job.

The Power Struggle

Powering 8 servos at once is no joke. If they all start moving simultaneously, they draw a massive current spike that browns out the ESP32 (basically rebooting the robot).

I solved this in two ways:

1. Hardware: We use a high-discharge 3S LiPo battery (450mAh) regulated down to 5V via a buck converter.

2. Software: I wrote a "stagger" into the firmware. Even when you tell the robot to stand up, the code introduces a 20ms delay between each motor activation. It’s hard to notice, but it prevents the voltage from crashing.

If you’re building one, you can still do the hand-wired method (I wrote a guide for it!), but I highly recommend grabbing the PCB files from the repo and ordering a batch.