Protecting Pulsed High-Power Lasers with Real-Time Image Classification


Protecting Pulsed High-Power Lasers with Real-Time Image Classification

Kelling, J.; Gebhardt, R.; Helbig, U.; Bock, S.; Schramm, U.; Juckeland, G.

Learn how to combine computer vision techniques and deep learning to improve the sensitivity of a real-time, GPU-powered safety system. In Petawatt laser systems, firing at 10Hz, suddenly appearing scatterers can damage components.
Damage(-spreading) can be avoided by suspending operation immediately on occurrence of such an event.

We present our approach for the automatic detection of critical failure states from intensity profiles of the laser beam. By Incorporating quick feature detection and learned heuristics for feature classification, both real-time constraints and limited available training data are accommodated. Localization of triggering feature is crucial for when the problem is located in non-sensitive sections and will not be removed from the beam in production.

  1. extended abstract:
High-power lasers are operated at our research center for investigations of exotic states of matter and medical applications, among others. This project to improve the automatic shutdown/interlock system of two lasers (one in operation, one currently under construction) has the goal of reducing the probability of, potentially expensive, damage-spreading scenarios, while at the same time avoiding false alarms at high sensitivity.
The project to be presented is currently in a proof-of-concept phase, with workable proof existing for a specific failure mode for which data was available (the breaking of a single mirror). Next to the 100ms real-time constraint, the lack sufficient training data, demanded a two-stage approach to solve this problem: Classical feature detection with a low threshold works as a fast anomaly detector, followed by feature-classification using CNNs (mostly GoogLeNet) to identify true positive triggers.
From this, the audience can learn how to design for short response times (to which end we employ Caffe, OpenCV on GPU and use C++ as main programming language). The application also demonstrates how prior domain knowledge and known algorithms can be combine with machine learning to create heuristics to fill in gaps.

Keywords: Image Classification; Caffe; automatik Laser-safety shutdown; GoogLeNet

  • Lecture (Conference)
    GTC 2018 Silicon Valley, 26.-29.03.2018, San Jose, CA, USA
  • Lecture (Conference)
    1st MLC Workshop, 15.05.2018, Dresden, Deutschland

Permalink: https://www.hzdr.de/publications/Publ-26608
Publ.-Id: 26608