Saved in:
Bibliographic Details
Main Authors: Kulkarni, Abhijeet M., Poulakakis, Ioannis, Huang, Guoquan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.22789
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914116194533376
author Kulkarni, Abhijeet M.
Poulakakis, Ioannis
Huang, Guoquan
author_facet Kulkarni, Abhijeet M.
Poulakakis, Ioannis
Huang, Guoquan
contents Accurate full-body motion prediction is essential for the safe, autonomous navigation of legged robots, enabling critical capabilities like limb-level collision checking in cluttered environments. Simplified kinematic models often fail to capture the complex, closed-loop dynamics of the robot and its low-level controller, limiting their predictions to simple planar motion. To address this, we present a learning-based observer-predictor framework that accurately predicts this motion. Our method features a neural observer with provable UUB guarantees that provides a reliable latent state estimate from a history of proprioceptive measurements. This stable estimate initializes a computationally efficient predictor, designed for the rapid, parallel evaluation of thousands of potential trajectories required by modern sampling-based planners. We validated the system by integrating our neural predictor into an MPPI-based planner on a Vision 60 quadruped. Hardware experiments successfully demonstrated effective, limb-aware motion planning in a challenging, narrow passage and over small objects, highlighting our system's ability to provide a robust foundation for high-performance, collision-aware planning on dynamic robotic platforms.
format Preprint
id arxiv_https___arxiv_org_abs_2510_22789
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning Neural Observer-Predictor Models for Limb-level Sampling-based Locomotion Planning
Kulkarni, Abhijeet M.
Poulakakis, Ioannis
Huang, Guoquan
Robotics
Accurate full-body motion prediction is essential for the safe, autonomous navigation of legged robots, enabling critical capabilities like limb-level collision checking in cluttered environments. Simplified kinematic models often fail to capture the complex, closed-loop dynamics of the robot and its low-level controller, limiting their predictions to simple planar motion. To address this, we present a learning-based observer-predictor framework that accurately predicts this motion. Our method features a neural observer with provable UUB guarantees that provides a reliable latent state estimate from a history of proprioceptive measurements. This stable estimate initializes a computationally efficient predictor, designed for the rapid, parallel evaluation of thousands of potential trajectories required by modern sampling-based planners. We validated the system by integrating our neural predictor into an MPPI-based planner on a Vision 60 quadruped. Hardware experiments successfully demonstrated effective, limb-aware motion planning in a challenging, narrow passage and over small objects, highlighting our system's ability to provide a robust foundation for high-performance, collision-aware planning on dynamic robotic platforms.
title Learning Neural Observer-Predictor Models for Limb-level Sampling-based Locomotion Planning
topic Robotics
url https://arxiv.org/abs/2510.22789