You are here

Accelerated High-Power Blue Laser Design Cycle Enabled by Deep Neural Networks

Description:

OUSD (R&E) CRITICAL TECHNOLOGY AREA(S): Trusted AI and Autonomy; Microelectronics The technology within this topic is restricted under the International Traffic in Arms Regulation (ITAR), 22 CFR Parts 120-130, which controls the export and import of defense-related material and services, including export of sensitive technical data, or the Export Administration Regulation (EAR), 15 CFR Parts 730-774, which controls dual use items. Offerors must disclose any proposed use of foreign nationals (FNs), their country(ies) of origin, the type of visa or work permit possessed, and the statement of work (SOW) tasks intended for accomplishment by the FN(s) in accordance with the Announcement. Offerors are advised foreign nationals proposed to perform on this topic may be restricted due to the technical data under US Export Control Laws. OBJECTIVE: Develop an automated high-peak-power blue laser with high-repetition-rate design process via using neural networks and machine learning (ML) algorithms that will result in up to 50 times reduction in design cycle time compared to the conventional “manual” laser design process. DESCRIPTION: The Navy requires a high-peak-power blue laser system solution to be operated in pulsed mode with high-repetition rate for standoff oceanographic sensing applications from an aircraft at an altitude of under 1000 ft (304.8 m). It should be ruggedized and sufficiently small in size, weight, and power consumption (SWaP), to be used in naval fixed- and rotary-wing platforms. The current state of the art, which includes Optical Para-Metric Oscillators (OPOs), wavelength doubling of titanium-sapphire (TiSa) based lasers, doubling and tripling of other laser hosts, and blue laser diodes, do not adequately support the naval-demanding performance, size, and weight objectives. Many commercially available lasers and near-term developmental lasers meet a few of the required characteristics, but none can meet every performance criteria. It is paramount that the blue laser solution meets or exceeds the design objectives in order to be effective for standoff oceanographic sensing. The performance specification of this laser solution include, but are not limited to: (a) high-repetition rate (Threshold: 250 Hz), (b) high-peak-power (> 20 MJ per pulse with pulse width no more than 25 ns), (c) blue wavelength at 47X nm, (d) Spectral line width of less than or equal to 0.1 nm, (e) wall-plug efficiency of greater than 5%, (f) laser beam quality m² - 3, (g) lightweight. (Total weight including the laser head, cooling system, power supply, and control system) < 50 lb (22.68 kg), (h) small volume. (Total volume for the cooling system, power supply, control system and laser head) < 2 ft³ (.057 m³), (i) ability to be ruggedized and packaged to withstand the shock, vibration, pressure, temperature, humidity, electrical power conditions, and so forth, encountered in a system built for airborne use, (j) reliability: Mean time between equipment failure—300 operating hr. Whether the laser design is based on a multistage OPO architecture or frequency conversion of diode-pumped solid-state laser, the laser performance metrics (such as emission wavelength, threshold-current density, pulse width, repetition rate, slope efficiency, and their temperature dependence) are closely linked to the intricate interrelationship among the brightness of the pump laser. Suppression of the unintended parasitic solid-state laser emission, temperature control of the solid-state crystal, and the nonlinear properties of the nonlinear crystal, and so forth. The complexity of the architecture generally requires a time-consuming iterative process between experiment and design optimization to achieve the highest device performance, which adds substantial cost to laser manufacturing. Automated optimization algorithms similar to the one used in References 1 and 2 could both greatly reduce the time (and cost) required to develop new high peak power blue laser systems with specified performance characteristics and potentially lead to new insights into blue laser design. The current blue laser design process generally involves a human in the loop—even for a single iteration. The function performed by the human is to identify specific features in the design and determine whether a certain performance metric can be achieved. Emerging data-driven automated optimization algorithms could potentially address the difficulties facing this laser design. As the blue laser requirements grow and progress, the design processes become more challenging. With conventional design approaches based on computational optimization, one typically starts with a prior design and computes the performance, compared to the target response. The parameters in each of the active components in the multistage architecture are calculated and applied to the design. This process, performed repetitiously, often takes many iterations before a design is found that meets the design criteria. As an alternative, the data-driven approach [Refs 1 and 2] is rapidly emerging where deep neural networks are used for inverse device design. A large data set of existing designs and corresponding performances can be used to train artificial neural networks [Ref 3] so that the networks can develop intuitive connections between the laser system designs and their performances. After training, the neural network can accomplish a design goal in hours instead of weeks as compared to the conventional approach. Such an approach has been used previously in photonic structures [Ref 2], where neural networks successfully model the wave dynamics in the Maxwell’s equations and the quantum mechanics in the laser architectures. This SBIR topic seeks the development of a power scalable blue laser system solution that will meet the aforementioned size, weight, performance, and reliability requirements via a multiphysics-based, deep neural network, ML process. It is also the objective of this topic to accelerate the development cycle time of the laser aided by the ML compared to the conventional “manual” design process by at least a factor of 50. PHASE I: Develop a methodology for implementing the training plan for neural network-based blue laser design optimization without human intervention. Develop feasibility of a ML process, which is suitable for advancing the performance of the blue laser with respect to the performance specifications stated in the Description. The ML process should address all design parameters of the multistage laser architecture. Develop the design verification plan for the ML algorithm for accelerating the blue laser prototype development. The Phase I effort will include prototype plans to be developed in Phase II. PHASE II: Demonstrate the fully automated blue laser design algorithms using machine learning methodology. Perform experimental verification of the generated designs by demonstrating that the blue laser performance metrics are met with less than +/- 5% variations from the target performance specifications. Develop a prototype blue laser system based on the ML process that meets the required laser system specifications. Deliver the fully automated blue laser design algorithms with complete and detailed user manual and documentations. Benchmark the design cycle time using the algorithm aided by ML against the conventional method without using ML, and verify the cycle time reduction. PHASE III DUAL USE APPLICATIONS: Transition the technology for DoD use. Test and finalize the technology based on the design and simulation results developed during Phase II. Transition the design algorithm for DoD applications in the areas of standoff oceanographic sensing applications. Commercialize the design algorithm enabled by deep neutral networks from this effort for law enforcement, marine navigation, medical applications, and industrial manufacturing processing. REFERENCES: 1. Bismuto, A., Terazzi, R., Hinkov, B., Beck, M., & Faist, J. (2012). Fully automatized quantum cascade laser design by genetic optimization. Applied Physics Letters, 101(2), 021103. https://doi.org/10.1063/1.4734389 2. Liu, D., Tan, Y., Khoram, E., & Yu, Z. (2018). Training deep neural networks for the inverse design of nanophotonic structures. Department of Electrical and Computer Engineering, University of Wisconsin: Madison, WI. https://arxiv.org/ftp/arxiv/papers/1710/1710.04724.pdf 3. Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., van den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., Lanctot, M., Dieleman, S., Grewe, D., Nham, J., Kalchbrenner, N., Sutskever, I., Lillicrap, T., Leach, M., Kavukcuoglu, K., Graepel, T., & Hassabis, D. (2016). Mastering the game of Go with deep neural networks and tree search. Nature, 529(7587), 484-489. https://doi.org/10.1038/nature16961 KEYWORDS: Accelerated; High Power; Blue Laser; Design Cycle Time Reduction; Deep Neural Networks; Machine Learning
US Flag An Official Website of the United States Government