Proprioceptive State Estimation of Legged Robots with Kinematic Chain Modeling

TL;DR

This paper introduces a novel proprioceptive state estimation method for legged robots that leverages kinematic chain modeling within a factor graph framework, reducing reliance on visual data and improving accuracy.

Contribution

The work presents a new state estimation approach using kinematic chain models and factor graphs, outperforming existing methods and applicable across various legged robot platforms.

Findings

Outperforms current proprioceptive methods by 27% on average

Works effectively in simulation and hardware environments

Generalizes across different legged robot platforms

Abstract

Legged robot locomotion is a challenging task due to a myriad of sub-problems, such as the hybrid dynamics of foot contact and the effects of the desired gait on the terrain. Accurate and efficient state estimation of the floating base and the feet joints can help alleviate much of these issues by providing feedback information to robot controllers. Current state estimation methods are highly reliant on a conjunction of visual and inertial measurements to provide real-time estimates, thus being handicapped in perceptually poor environments. In this work, we show that by leveraging the kinematic chain model of the robot via a factor graph formulation, we can perform state estimation of the base and the leg joints using primarily proprioceptive inertial data. We perform state estimation using a combination of preintegrated IMU measurements, forward kinematic computations, and contact detections in a factor-graph based framework, allowing our state estimate to be constrained by the robot model. Experimental results in simulation and on hardware show that our approach out-performs current proprioceptive state estimation methods by 27% on average, while being generalizable to a variety of legged robot platforms. We demonstrate our results both quantitatively and qualitatively on a wide variety of trajectories.