Bird's-eye View Image Generation with Camera Malfunction in Irradiation Environment

Ren Komatsu, Hiromitsu Fujii, Hitoshi Kono, Yusuke Tamura, Atsushi Yamashita, Hajime AsamaThe Abstracts of the International Conference on Advanced Mechatronics Toward Evolutionary Fusion of IT and Mechatronics ICAM 2015 · DOI

Abstract

In this study, a gamma irradiation experiment for our proposed bird's-eye view system was conducted. Experimental results show that cameras attached to our system malfunctioned one by one in the irradiation environment, and the proposed system generated bird's-eye view images until all cameras malfunctioned.

Introduction

Teleoperating robots in extreme environments, such as the Fukushima Daiichi nuclear plant, presents significant spatial awareness challenges for operators. A bird’s-eye view (BEV)—a synthesized 360-degree top-down perspective—is often essential for effective navigation. However, high-dose gamma radiation in these zones typically causes image sensors to fail rapidly. Standard BEV systems are fragile; the loss of a single camera creates blind spots that compromise the entire navigation interface. This research proposes a fault-tolerant BEV system designed to maintain operational visibility through graceful degradation as individual hardware components fail.

Overview of the fault-tolerant bird's-eye view (BEV) system, showing the conversion from fish-eye feeds to a synthesized top-down perspective for teleoperation.

Methodology

The system utilizes four fish-eye cameras (AXIS M3007-PV) mounted on a 600mm x 600mm frame, simulating a robotic platform. The cameras are positioned to ensure that every floor region is covered by at least two sensors, providing necessary redundancy. A synthesis algorithm detects camera malfunctions and re-stitches the BEV image using only the remaining functional feeds.

To validate the system, the setup was exposed to a Cobalt-60 (Co-60) radiation source at the ATOX Co., Ltd. Engineering Research & Development Center. The experiment tracked the time to malfunction and total absorbed dose for each camera relative to its position.

Experimental layout illustrating the 600mm square camera configuration and its proximity to the radiation source.

Results and Discussion

The cameras exhibited staggered failure, succumbing to radiation at total doses of 141.3 Gy, 162.9 Gy, 192.1 Gy, and 224.1 Gy. This staggered failure is attributed to varying distances from the radiation source and inherent hardware tolerances.

Crucially, the system successfully maintained a synthesized BEV image even with only one functional camera remaining. While "salt-and-pepper" noise increased proportionally with the radiation dose, the system provided continuous, albeit degraded, visual feedback for 23 to 94 minutes depending on the local dose rate (ranging from 104 to 501 Gy/h). This proves that redundancy is a viable strategy for extending mission life in lethal environments.

Conclusion

The proposed redundant BEV framework demonstrates significant resilience in high-radiation environments. By prioritizing operational continuity over perfect image quality, the system extends the functional life of teleoperated robots in disaster zones. Future work will focus on real-time noise reduction filters to mitigate radiation-induced sensor artifacts and improve visibility for human operators during the final stages of sensor life.

Summary updated

Abstract

In this study, a gamma irradiation experiment for our proposed bird's-eye view system was conducted. Experimental results show that cameras attached to our system malfunctioned one by one in the irradiation environment, and the proposed system generated bird's-eye view images until all cameras malfunctioned.
BibTeX
@inproceedings{komatsu2015bird,
  author = {Ren Komatsu and Hiromitsu Fujii and Hitoshi Kono and Yusuke Tamura and Atsushi Yamashita and Hajime Asama},
  title = {{Bird's-eye View Image Generation with Camera Malfunction in Irradiation Environment}},
  booktitle = {The Abstracts of the International Conference on Advanced Mechatronics Toward Evolutionary Fusion of IT and Mechatronics ICAM},
  year = {2015},
  doi = {10.1299/jsmeicam.2015.6.177},
}
Figure