Learning safer walking
Step-to-step prediction informs reinforcement learning for bipedal locomotion.
Watch & exploreResearch in motion
Video Gallery
Research demonstrations and supplementary paper videos.
Follow the Cyberbotics Lab YouTube channel for the latest uploads.
20 videos · newest first
Walking across disconnected footholds
Perception and variable step timing for walking across disconnected footholds.
Watch & explore
Learning navigation from prior data
Data bootstrapping supports reinforcement learning for dynamic navigation subgoals.
Watch & explore
Navigation with control barrier functions
Linear discrete control barrier functions for safe bipedal robot navigation.
Watch & explore
Learning compact walking representations
A data-driven latent representation supports bipedal locomotion learning.
Watch & explore
Learning with simplified walking models
Template models inform task-space learning for bipedal locomotion.
Watch & explore
Testing disturbance rejection
Disturbance-rejection testing of legged locomotion with a linear impactor.
Watch & explore
Navigating a changing environment
Real-time bipedal robot navigation in dynamic environments.
Watch & explore
Planning paths around obstacles
Control barrier functions and logistic regression for planning around polynomial-shaped obstacles.
Watch & explore
Learning robust walking
Reinforcement learning and cascade policies for robust bipedal locomotion.
Watch & explore
Adapting bipedal motion control
Resolved motion control using a linear inverted pendulum model and neural adaptation.
Watch & explore
Walking over rough terrain
Learned feedback motion policies for Digit walking on rough terrain.
Watch & explore
Digit’s first walking experiments
Digit’s first walking gait and small disturbance-rejection tests in the Cyberbotics Lab.
Watch & explore
First steps with the OSU controller
Early exoskeleton walking with the Ohio State controller.
Watch & explore
Learning with hybrid zero dynamics
Hybrid zero dynamics informs reinforcement learning for 3D bipedal walking.
Watch & explore