Robotics & hardware ·
The first real-world jump test of a homemade robot, trained entirely in simulation.
The builder trained a reinforcement learning policy in sim and loaded it straight onto the physical robot. On this attempt it pushed past its own torque and velocity limits and shut itself down mid-jump.
This is the hard part of robotics right now - the "sim-to-real" gap, where a policy that looks perfect in a simulator meets friction, backlash, and real motors.
Made by u/Ok_Stress3654. This write-up follows the original post; it is not an independent product test.
Keep up with what gets built.
Read the daily AI briefing, browse the tools, or sign up for briefing email updates.