A Skyentific host describes building and wiring a 16-kilogram bipedal robot from the company’s modified X825 and X1040 actuators, then tests its motor power with help from a Jetson Orin Nano.
The project moves from an Isaac Lab simulation to a physical bipedal robot. Each leg has five degrees of freedom driven by two X825 and three modified X1040 actuators. The robot weighs about 16 kilograms and carries two batteries, a Jetson Orin Nano, two USB-to-CAN adapters, and an IMU.
The electronics mount holds the battery management board, CAN adapters, IMU, and Jetson. After routing and securing cables, the host calibrates the actuators, adds capacitors across each joint’s power supply, and completes the wiring. The Jetson uses an SSD and a headless setup, with SSH access over USB-C and file transfers via SCP; the IMU worked over I²C after a two-day attempt to connect it over SPI.
The robot’s PLA feet are slippery, so the host uses a rubbery yoga mat for traction while planning better feet later. A pulley and belt rig supports the robot during initial tests, letting the host run Python programs on the Jetson over SSH without holding the robot by hand.
The host runs programs for basic motor movements, sitting and standing, and a one-legged sitting pose. The robot completes the sit-to-stand tests, including the one-legged version, while the host holds it because it has no balance algorithm yet. The results suggest the motors have enough power for walking, which the host plans to test with reinforcement learning in a later video.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.
Skyentific’s presenter explains how they built and trained a bipedal robot with NVIDIA Isaac Sim and Isaac Lab, transferred its policy to hardware, and explored Isaac Lab Arena.
