← Back

PiCrawler

A 4-legged walking robot that always reminded me of a crab. This project was great practice for building an end-to-end robotics project. I used a kit version to start then designed a stronger model and used Isaac Sim for the first time to simulate and iterate, although, ROS practice washe main goal of this project.

ROS Legged Locomotion Simulation End-to-End Robotics

Overview

Crab-like walker

PiCrawler is a compact quadruped walker designed to explore legged locomotion and full-stack robotics development. The project started with a kit-based build to get real motion quickly, then evolved into a redesigned CAD model focused on cleaner packaging and better geometry for simulation and iteration. Along the way, it served as an initial entry point into Isaac Sim while building practical experience using ROS as the integration layer.

Real kit version

Build + run

Hands-on robot experience

The first version was built from a kit and tested in the real world. It was fun and was a way to get the ball rolling quickly while helping build the initial understanding. Getting a real system walking quickly made it much easier to understand what I could improve and how I could make it cooler.

PiCrawler kit-built real robot walking platform
Kit-built PiCrawler: the version used for initial testing and learning.

Redesign + Isaac Sim

Better model, better iteration

PiCrawler redesigned CAD model used for Isaac Sim
Redesigned PiCrawler model: improved geometry and structure, used for simulation in Isaac Sim.

Simulation-first development

I designed a cleaner, more robust model and used it for simulation in Isaac Sim. This let me sanity-check motion, explore gaits, and iterate on geometry without risking hardware. It also became a strong first step into the Isaac Sim ecosystem, learning how to set up scenes, tune physics, and validate behavior.

ROS end-to-end practice

Integration layer

PiCrawler was a great end-to-end robotics exercise: hardware brings real constraints, simulation enables fast iteration, and ROS provides the structure to connect control, messaging, and experimentation into one workflow. It helped build comfort with the “full robotics stack” mentality. Individual components and specializations are cool, but what makes robotics the best in the world is making it all work together to bring an idea into reality.