Humans as Path-Finders for Safe Navigation

TL;DR

This paper proposes a three phase approach to fulfil the goal of identification and tracking of the person in the image space, sensor fusion between camera data and laser sensors, and point interpolation with continuous curvature curves.

Abstract

One of the most important barriers toward a widespread use of mobile robots in unstructured and human populated work environments is the ability to plan a safe path. In this paper, we propose to delegate this activity to a human operator that walks in front of the robot marking with her/his footsteps the path to be followed. The implementation of this approach requires a high degree of robustness in locating the specific person to be followed (the leader). We propose a three phase approach to fulfil this goal: 1. identification and tracking of the person in the image space, 2. sensor fusion between camera data and laser sensors, 3. point interpolation with continuous curvature curves. The approach is described in the paper and extensively validated with experimental results.

Figures (7)

• Figure 1: Overall scheme of the algorithm.
• Figure 2: (a) Robot sensing system setup, consisting of LIDAR sensor, RealSense D435, and RealSense T265 (for the visual odometry). (b) Laser scanned cloud points (thin black dots), with the object centroids (thick blue points) expressed in the LIDAR reference system $\langle L\rangle$.
• Figure 3: Two examples of the application of the drift tolerance \ref{['eq:drift']}. In (a), the person is correctly classified as a mismatch because the bounding box is too far apart from the region defined by \ref{['eq:drift']}, while in (b) the leader is newly detected as expected.
• Figure 4: (a) Initialisation phase: the detected leader is depicted with a blue rectangle. (b) Following phase: the leader is correctly recognised (green rectangle), while another person is a negative sample (red rectangle).
• Figure 5: Example of path fitting and reconstruction. The red stars represent the input data. The green squares are the fitted waypoints, sampled at a uniform distance along the path. The blue solid line is the reconstructed, smoothed path, to be followed by the robot.