Preprint
Aug 2026
TRACE: Learned Proprioceptive Odometry for Legged Robots under Unreliable Contact Conditions
An end-to-end learned proprioceptive odometry estimator for legged robots under unreliable contact conditions that directly predicts relative displacement, relative rotation, and body-frame velocity from a recent history of onboard inertial and joint measurements.
Taehyeon Kong, W. Kim, Jemin Hwangbo
· 0 citations