Third-version biped robot with smoother walking and single-leg hops

Oct 27, 2026 Last reply: 12 hours ago 4 Replies
Jump to replies (4)

Most hobby bipeds I see either shuffle with fixed servo sweeps or never leave the safety rig. This third iteration of a DIY biped does both: it walks with a noticeably smoother gait and it even tries hopping on one leg. Compared with the jittery first version, the V3 walks with more confidence and far less shake, a real jump in quality for a home build.



The mechanical layout looks thought through. A blue and white chassis hides the internal wiring, longer pale leg linkages connect to the blue panels, and modular actuators sit at the hip joints, with Gim6010-8 24v motors providing the power. The robot runs tethered to a long horizontal boom arm on a weighted central pivot; in my view that is a sensible way to test a walker that still tips over. Driving is handled with a black wireless gamepad, while a laptop displays telemetry with gait trajectory graphs and a kinematic stick figure.



Out on the floor it walks across hardwood and foam mats, takes alternating steps forward and reverses along a circular path. During hopping tests the single leg pushes off the yoga mats in rapid cycles. Single-leg balancing is nailed in simulation, but in the real hopping test the robot lost balance and tipped forward. Closing that sim-to-hardware gap is the open question here, in my view.



Durability is the sticking point. A joint broke three times during testing, and a GT2 belt let go under tension partway along a straight path. There is also a harsh moment when a stop command follows a forward command. With the maker already considering future jumps and even a backflip, sorting those weak points first seems like the wiser route to me.


Mounting the actuators right at the hip joints looks like a clever call to me. In my view keeping the drive mass close to the joint helps a gait stay smooth, and the Gim6010-8 24v motors seem a solid pick for a frame this size. The trade-off as I see it is maintenance: any repair likely means working deep inside the blue and white chassis.

I am curious about the hopping test. When the robot tipped forward during the single-leg attempt, was the boom arm still holding it, or does the tether allow enough freedom that the leg has to catch the full body weight on its own?

The maker has not said whether the boom was bearing load at that moment. What is known is that the single leg pushed off the yoga mats in rapid cycles during the hopping tests, and that balancing on one leg works in simulation even though the real robot tipped forward.

Before chasing the backflip the maker is considering, I would reinforce the joints and add some way to watch belt tension, since one joint broke three times and a GT2 belt let go partway through a straight run. Getting the real-world single-leg balance to match the simulation result seems like the better next milestone to me.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required