An experiment on squad navigation of human and robots

IARP/EURON Workshop on Robotics for Risky Interventions and Environmental Surveillance

January 7th-8th, 2008 - Benicàssim (Spain)

L. Nomdedeu, J. Sales, E. Cervera, J. Alemany,R. Sebastia, J. Penders, V. Gazi

Overview

• Introduction

• System configuration

• Implementation

• The experiment

• Conclusion and future work

Introduction: motivation

• Experiment: a set of robots (acting as a swarm) and one person.

• The aim of robot-human squad: to explore an indoors area.

• Fire-fighters: – no sensing capabilities– no map of the building

Introduction: statement of the problem

• Our system must be able to integrate:– Exploration and obstacle avoidance– Following the human and localizing itself

• Pose problem:– Tracking the human with the escorts lasers– Providing him with an accelerometer-

gyroscope-magnetometer (IMU)

Introduction: related work

• Laser range finders:– To find and track humans– Detecting multiple moving objects, multiple humans

• Video data:– lack of visibility

• Localization, obstacle avoidance, path-planning– Adaptive Monte Carlo Localization Plus– Vector Field Histogram (VFH)– Nearness Diagram (ND)

System configuration: HW setup

• Robots: mobile commercial platform Erratic from Videre-design– lasers readings, – encoder data,– sonar data,– imu data

System configuration: HW setup

• Fire-fighter: laptop along with an imu sensor

System configuration: SW setup

• Laptop will provide the fire-fighter – an estimation of his own pose– a new orientation and velocity to take

• Three roles in our

human-robot system

System configuration: SW setup

• Logical scheme of the system

Implementation: SW architecture

• Hardware Abstraction Layer (HAL) provided by the Player platform.

Implementation: tracking

• Human being localization through only the laser rangefinder data

(,r): polar coordinates of the fire-fighter centroid

: threshold of distance

: angle of the reading

Implementation: tracking

Implementation: tracking

• Estimated global position and velocity of the fire-fighter: – Relative fire-fighter position and velocity– Robot’s own estimated position

and velocity.

• Formation rules:

The experiment

• Partially tested in simulation• Robots guiding the

human through our department

• Human: only laptop information

• Collision free path• Occlusions along

the path

Conclusion and future work

• We have validated the use of mobile platforms in hazardous environments as a support for human beings

• More robustness can be implemented:– Supporting more unstructured environments– Improving system reliability– Improving pose fusion and prediction– Online SLAM approach– Full Human Augmented Mapping (HAM) system