Q8botOne: Small Robot. Big Possibilities.

Some robotics projects begin with a big lab. This one can begin in the palm of your hand.

Meet Q8botOne, the next generation of Yufeng Wu’s Q8bot project from ZeroWire Robotics. Its Crowd Supply campaign launched on October 6, 2026, bringing a refined platform to makers, educators, and robotics experimenters. The launch announcement highlights jogging, jumping, payload carrying, custom gaits, and expressive motions. Think of it as a tiny moving canvas for your next idea.

For ROBOTIS, the connection is especially exciting: DYNAMIXEL is at the heart of its motion.

Small hardware. Room to experiment.

Q8botOne’s campaign specifies eight DYNAMIXEL XL330-M077-T smart actuators, four parallel-linkage legs, and a central PCB that replaces conventional internal cabling. Digital joint communication opens access to position, velocity, and tunable control parameters. Explore the hardware specifications.

That makes the robot an appealing place to ask hands-on questions. What happens when the feet spend longer on the ground? How does a shorter step look at a higher cadence? Can a crouch, pause, and little hop make a machine seem curious? Start with one motion, change one variable, and watch the result.

Can you retrain its gait?

You can program and tune custom gaits. That capability is explicitly described in the Q8botOne launch update. Machine-learning “retraining” is a different proposition.

The original Q8bot design paper explains precomputed foot trajectories and phase offsets for different gaits. In its future-control discussion, reinforcement-learning policies are a research direction that would benefit from state estimation, additional sensing, and potentially upgraded compute.

Your first experiment starts here.

Drive a real CAD-derived Q8bot model in your browser, powered by MuJoCo physics. Try the creator’s official walk and trot trajectories, orbit the robot to inspect its five-bar legs, and experiment with floor grip, a gentle ramp, or miniature steps. Suspend the robot to watch each linkage at work.

This lab uses the original Q8bot HW v2.5 CAD assembly and the creator’s published gait code, with actual MuJoCo contact dynamics. Q8botOne is a newer product; this is not its verified digital twin. Servo dynamics, inertias, and collision surfaces are estimates. The lab does not train an AI policy or establish hardware performance.

What would you build?

1. A robot with a personality. Choreograph a greeting, a curious lean, or a tiny victory dance. Treat motion like animation: timing and pauses matter as much as the pose.
2. A gait comparison bench. Try different foot trajectories and leg timing. On real hardware, record repeatable trials rather than assuming the most dramatic-looking gait is the best one.
3. A sensor-driven explorer. The campaign lists Qwiic/STEMMA QT sensor expansion and powered UART for additional compute. A distance sensor or camera could become the starting point for a custom obstacle-response project. The sensor, code, and integration would be your additions. See expansion details.
4. A classroom motion challenge. Give each team the same starting gait and a different goal: a precise turn, an expressive gesture, or a repeatable route. Ask students to explain what they changed and why.
5. Your own hardware variation. The original Q8bot repository shares hardware resources, assembly instructions, firmware, and Python tools. It is a useful design reference; choose files and software that match your robot’s revision.

From an idea on screen to a robot on your desk.

Q8botOne builds on an independently developed open-source project. The campaign describes refinements including built-in charging, a dedicated controller, and easier assembly. It is a crowdfunding launch; visit the campaign for current production status and delivery estimates. Read the creator’s launch story.

The kit and assembled robot are different packages: the campaign’s kit excludes the eight DYNAMIXEL actuators. Check the current contents before choosing your starting point. Compare campaign options.

Created by Yufeng (Eric) Wu / ZeroWire Robotics. Original Q8bot research by Yufeng Wu and Dennis Hong. ROBOTIS highlights the project as part of the DYNAMIXEL ecosystem. Product details checked October 9, 2026.