Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

FPGA-accelerated adaptive optics wavefront control

 
: Mauch, Steffen; Reger, Johann; Reinlein, Claudia; Appelfelder, Michael; Goy, Matthias; Beckert, Erik; Tünnermann, Andreas

:

Bifano, T.G. ; Society of Photo-Optical Instrumentation Engineers -SPIE-, Bellingham/Wash.:
MEMS Adaptive Optics VIII : February 2014, San Francisco, California, United States
Bellingham, WA: SPIE, 2014 (SPIE Proceedings 8978)
ISBN: 978-0-8194-9891-5
Paper 897802
Conference "MEMS Adaptive Optics" <8, 2014, San Francisco/Calif.>
Englisch
Konferenzbeitrag
Fraunhofer IOF ()
adaptive optics; PCIe; FPGA; deformable mirror; laser material processing; Linux real-time system; wavefront sensor

Abstract
The speed of real-time adaptive optical systems is primarily restricted by the data processing hardware and computational aspects. Furthermore, the application of mirror layouts with increasing numbers of actuators reduces the bandwidth (speed) of the system and, thus, the number of applicable control algorithms. This burden turns out a key-impediment for deformable mirrors with continuous mirror surface and highly coupled actuator influence functions. In this regard, specialized hardware is necessary for high performance real-time control applications. Our approach to overcome this challenge is an adaptive optics system based on a Shack-Hartmann wavefront sensor (SHWFS) with a CameraLink interface. The data processing is based on a high performance Intel Core i7 Quadcore hard real-time Linux system. Employing a Xilinx Kintex-7 FPGA, an own developed PCie card is outlined in order to accelerate the analysis of a Shack-Hartmann Wavefront Sensor. A recently developed real-time capable spot detection algorithm evaluates the wavefront. The main features of the presented system are the reduction of latency and the acceleration of computation For example, matrix multiplications which in general are of complexity O(n3 are accelerated by using the DSP48 slices of the field-programmable gate array (FPGA) as well as a novel hardware implementation of the SHWFS algorithm. Further benefits are the Streaming SIMD Extensions (SSE) which intensively use the parallelization capability of the processor for further reducing the latency and increasing the bandwidth of the closed-loop. Due to this approach, up to 64 actuators of a deformable mirror can be handled and controlled without noticeable restriction from computational burdens.

: http://publica.fraunhofer.de/dokumente/N-283239.html