[Paper Review] An Open Source, Fiducial Based, Visual-Inertial State Estimation System.

Michael Neunert, Michael Blösch+1 more|arXiv (Cornell University)|Jul 8, 2015

Robotics and Sensor-Based Localization6 references3 citations

TL;DR

This paper presents an open-source, fiducial-based visual-inertial state estimation system that uses printable paper markers to enable accurate, low-cost localization and motion tracking in indoor and constrained environments. By fusing visual data from fiducials with inertial measurements, the system achieves robust, continuous pose estimation without requiring additional calibration, even during fast motions.

ABSTRACT

Many robotic tasks rely on the estimation of the location of moving bodies with respect to the robotic workspace. This information about the robots pose and velocities is usually either directly used for localization and control or utilized for verification. Often motion capture systems are used to obtain such a state estimation. However, these systems are very costly and limited in terms of workspace size and outdoor usage. Therefore, we propose a lightweight and easy to use, visual inertial Simultaneous Localization and Mapping approach that leverages paper printable artificial landmarks, so called fiducials. Results show that by fusing visual and inertial data, the system provides accurate estimates and is robust against fast motions. Continuous estimation of the fiducials within the workspace ensures accuracy and avoids additional calibration. By providing an open source implementation and various datasets including ground truth information, we enable other community members to run, test, modify and extend the system using datasets or their own robotic setups.

Motivation & Objective

• To develop a low-cost, lightweight alternative to expensive motion capture systems for robotic state estimation.
• To address the limitations of motion capture systems, such as high cost, restricted workspace size, and poor outdoor usability.
• To enable accurate and robust pose estimation using only visual and inertial sensors with printable fiducial markers.
• To provide an open-source implementation with datasets and ground truth for community testing, modification, and extension.

Proposed method

• The system uses paper-printable artificial landmarks (fiducials) as visual references in the environment.
• It fuses visual observations of fiducials with inertial measurements from an IMU to estimate the robot's state.
• A state estimation filter combines visual and inertial data to produce continuous, accurate pose and velocity estimates.
• The system avoids additional calibration by continuously tracking fiducials within the workspace.
• The implementation is open source, enabling integration with various robotic platforms and datasets.
• Ground truth data is provided in the form of datasets to validate and benchmark performance.

Experimental results

Research questions

• RQ1 Can a low-cost, fiducial-based visual-inertial system achieve accurate and robust state estimation comparable to motion capture systems?
• RQ2 How well does the fusion of visual and inertial data perform under fast motion conditions?
• RQ3 To what extent can continuous fiducial tracking eliminate the need for external calibration?
• RQ4 How scalable and reusable is the open-source system across different robotic platforms and environments?

Key findings

• The system achieves accurate state estimation by effectively fusing visual and inertial data, even during fast motions.
• The use of printable fiducials enables low-cost deployment without requiring expensive infrastructure.
• Continuous fiducial tracking ensures consistent accuracy and eliminates the need for additional calibration procedures.
• The open-source implementation with provided datasets allows for reproducible testing and extension by the research community.
• The system demonstrates robustness in real-world conditions, offering a viable alternative to traditional motion capture systems.