Autonomous Control of Airship for Indoor Experiment

Ikiyoshi Tomita, Yasushi Hada, Y. Fukuda, Kimihisa Kawabata, Hayato Kaetsu, Hajime Asama, Y. KurodaThe Proceedings of JSME Annual Conference on Robotics and Mechatronics (Robomec) 2004 · DOI

Abstract

In the aftermath of disasters such as earthquakes, indoor reconnaissance is critical yet hazardous for human rescuers. This paper explores the development of an autonomous airship designed for indoor search-and-rescue operations. Given the GPS-denied nature of indoor environments, the study evaluates the efficacy of onboard inertial sensors compared to external optical tracking systems. Experimental results identify a critical 10-second threshold for reliable position estimation using onboard accelerometers, highlighting the necessity of hybrid localization strategies for sustained autonomous flight.

Introduction

The use of aerial robotics in disaster response offers a significant safety advantage, allowing for the inspection of structurally compromised buildings without risking human lives. Among various platforms, small airships are particularly well-suited for indoor environments. Their inherent buoyancy makes them safer for operation around people and obstacles, and their ability to hover for extended periods with minimal power consumption is ideal for detailed search missions.

However, achieving full autonomy in these environments presents a significant challenge: localization. Without access to Global Positioning System (GPS) signals, an airship must rely on alternative methods to determine its 3D coordinates. This research investigates the limitations of dead reckoning using onboard sensors and the role of external ground-based tracking.

Methodology

The researchers developed a prototype airship system incorporating two distinct sensing modalities to evaluate localization accuracy:

  1. Internal Sensing (Dead Reckoning): The airship is equipped with onboard accelerometers. By mathematically integrating acceleration data over time—a process known as dead reckoning—the system attempts to estimate its current velocity and position.
  2. External Sensing (Ground Truth): To provide a high-precision reference, a "Total Station" (an optical surveying instrument) was deployed on the ground. This device tracks the airship's position in real-time, providing a global coordinate system that serves as the "ground truth" for the experiment.

The experimental flight maneuvers were conducted indoors, comparing the position estimated by the onboard sensors against the precise tracking data provided by the Total Station.

Results and Discussion

The study's primary finding is the identification of a "10-second window" for inertial navigation. While the onboard accelerometers provided sufficient data for short-term stability and immediate reactive maneuvers, the accuracy of the position estimation degraded rapidly.

The Integration Trap and Sensor Drift

The degradation in accuracy is attributed to the "integration trap." Because position is derived from the double-integration of acceleration data, even infinitesimal measurement errors or sensor noise are compounded exponentially over time. This cumulative error, or "drift," results in the airship's estimated position diverging significantly from its actual location within seconds. The data confirmed that after approximately 10 seconds of flight, the airship could no longer reliably navigate based solely on internal inertial data.

The Role of the Total Station

The Total Station proved essential for maintaining spatial awareness over longer durations. By mapping the airship's flight path onto a global coordinate plot, the external system provided the necessary "anchor" to correct the drift inherent in the onboard sensors. This confirms that while internal sensors are useful for high-frequency stability, they cannot currently support long-term autonomous missions in isolation.

Conclusion and Future Work

This research demonstrates that for an airship to perform meaningful search-and-rescue tasks—which typically exceed the 10-second accuracy window of basic inertial sensors—a hybrid localization approach is mandatory.

Future developments in this project will likely focus on reducing the reliance on external hardware like the Total Station, which may be difficult to deploy in actual disaster zones. Key areas for future research include:

  • Sensor Fusion: Combining accelerometers with gyroscopes, ultrasonic sensors, and barometers to extend the duration of accurate dead reckoning.
  • Onboard Computer Vision: Implementing Simultaneous Localization and Mapping (SLAM) algorithms, allowing the airship to recognize environmental landmarks and "reset" its position internally.
  • System Miniaturization: Optimizing the weight of control electronics to maintain the airship's payload capacity for search equipment.

By overcoming these localization hurdles, autonomous airships can become a vital tool in the rapid assessment of indoor disaster environments.

Summary updated

BibTeX
@inproceedings{tomita2004autonomous,
  author = {Ikiyoshi TOMITA and Yasushi Hada and Y. Fukuda and Kimihisa Kawabata and Hayato Kaetsu and Hajime Asama and Y. Kuroda},
  title = {{Autonomous Control of Airship for Indoor Experiment}},
  booktitle = {The Proceedings of JSME Annual Conference on Robotics and Mechatronics (Robomec)},
  year = {2004},
  doi = {10.1299/jsmermd.2004.40\_2},
}
Figure