Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Leg Exoskeleton Odometry using a Limited FOV Depth Sensor

Fabio Elnecave Xavier, Matis Viozelange, Guillaume Burger, Marine Pétriaux, Jean-Emmanuel Deschaud, François Goulette

TL;DR

This work tackles odometry and terrain perception for leg exoskeletons constrained by a limited FOV depth sensor. It introduces a point-cloud-to-elevation-map ICP that aligns depth data to a discretized elevation map and derives a covariance model to handle underconstrained geometry, integrated with a proprioceptive EKF to produce accurate trajectory estimates and elevation maps. Compared to a purely proprioceptive baseline and a point-cloud-map ICP variant, the elevation-map ICP approach yields reduced drift, smoother terrain maps, and faster computation, with experiments showing improved XY accuracy and robust performance through steps and doorways. The method has practical impact for real-world mobility of exoskeleton users, enabling safer navigation in constrained environments.

Abstract

For leg exoskeletons to operate effectively in real-world environments, they must be able to perceive and understand the terrain around them. However, unlike other legged robots, exoskeletons face specific constraints on where depth sensors can be mounted due to the presence of a human user. These constraints lead to a limited Field Of View (FOV) and greater sensor motion, making odometry particularly challenging. To address this, we propose a novel odometry algorithm that integrates proprioceptive data from the exoskeleton with point clouds from a depth camera to produce accurate elevation maps despite these limitations. Our method builds on an extended Kalman filter (EKF) to fuse kinematic and inertial measurements, while incorporating a tailored iterative closest point (ICP) algorithm to register new point clouds with the elevation map. Experimental validation with a leg exoskeleton demonstrates that our approach reduces drift and enhances the quality of elevation maps compared to a purely proprioceptive baseline, while also outperforming a more traditional point cloud map-based variant.

Leg Exoskeleton Odometry using a Limited FOV Depth Sensor

TL;DR

This work tackles odometry and terrain perception for leg exoskeletons constrained by a limited FOV depth sensor. It introduces a point-cloud-to-elevation-map ICP that aligns depth data to a discretized elevation map and derives a covariance model to handle underconstrained geometry, integrated with a proprioceptive EKF to produce accurate trajectory estimates and elevation maps. Compared to a purely proprioceptive baseline and a point-cloud-map ICP variant, the elevation-map ICP approach yields reduced drift, smoother terrain maps, and faster computation, with experiments showing improved XY accuracy and robust performance through steps and doorways. The method has practical impact for real-world mobility of exoskeleton users, enabling safer navigation in constrained environments.

Abstract

For leg exoskeletons to operate effectively in real-world environments, they must be able to perceive and understand the terrain around them. However, unlike other legged robots, exoskeletons face specific constraints on where depth sensors can be mounted due to the presence of a human user. These constraints lead to a limited Field Of View (FOV) and greater sensor motion, making odometry particularly challenging. To address this, we propose a novel odometry algorithm that integrates proprioceptive data from the exoskeleton with point clouds from a depth camera to produce accurate elevation maps despite these limitations. Our method builds on an extended Kalman filter (EKF) to fuse kinematic and inertial measurements, while incorporating a tailored iterative closest point (ICP) algorithm to register new point clouds with the elevation map. Experimental validation with a leg exoskeleton demonstrates that our approach reduces drift and enhances the quality of elevation maps compared to a purely proprioceptive baseline, while also outperforming a more traditional point cloud map-based variant.

Paper Structure

This paper contains 18 sections, 16 equations, 4 figures, 1 table.

Table of Contents

  1. INTRODUCTION
  2. POINT-CLOUD-TO-ELEVATION-MAP ICP
  3. Computing the rigid transformation
  4. Point cloud downsampling
  5. Correspondence estimation
  6. Normal vector computation
  7. Robust optimization
  8. Estimating the covariance matrix
  9. ODOMETRY AND MAPPING
  10. Updating the state estimator
  11. Updating the elevation map
  12. EXPERIMENTAL VALIDATION
  13. Methodology
  14. Results
  15. Experiment 1 -- climbing and descending a step
  16. ...and 3 more sections

Figures (4)

  • Figure 1: On the left, photo of the personal exoskeleton from Wandercraft with a depth camera mounted above its right knee. On the right, 3D view showing the limited FOV of the camera at this position as the exoskeleton climbs a step.
  • Figure 2: Trajectories from experiment 1. The rectangle in the XY plot indicates the location of the box. The starting point of the trajectory is marked with an "x".
  • Figure 3: Trajectories from experiment 2. The XY plot is overlaid on a floor plan of the building. The starting point of the trajectory is marked with an "x".
  • Figure : EKF-P