In a world of significant technological disruption, Defence’s ability to mitigate future threats is underpinned by a strong science and technology base. I will provide examples of how we are working across UK academia and industry to develop and exploit key sensing and imaging technologies that will deliver future operational advantage to UK forces.

Our interests range from understanding fundamental novel optical and quantum phenomena to working in partnership with universities and industrial partners to demonstrate key concepts. I'll provide examples from the area of quantum sensing, where there is a real drive to test prototype technologies in relevant environments on airborne and maritime platforms. At a systems level I'll talk about how MOD is leading the development of architectures like Sapient - a key enabler for future integrated ISTAR systems. As we seek to continually adapt to new opportunities, I'll also talk about emerging themes such as distributed coherent sensing.

I’ll close on one of the most important assets for the future of UK defence - a skilled UK workforce which we support through the funding of PhDs, including the recently announced the Sensing, Processing and AI for Defence (SPADS) Centre for Doctoral Training, jointly with EPSRC. Through such collaborations and our work with UK industry we are ensuring the UK armed forces can remain at the cutting edge of science and technology.”

Photon counting capable sensors tend to operate in a regime known as Deep Sub-Electron Read Noise or DSERN. In this regime, discrete nature of individual photons remains in the observed data, enabling quantification and new quantum experimental applications. A critical component to accurate photon counting is the knowledge (at the pixel level) of the conversion gain and read noise. This work compares the use of the Photon Counting Histogram Expectation Maximization (PCH-EM) algorithm to other DSERN characterization methods focusing on key performance parameters of conversion gain and read noise. A sensitivity analysis using synthetic data explores the dependence of the uncertainty in the conversion gain estimate on the magnitude of read noise and relative illumination levels. Additionally, the PCH-EM approach is validated using experimental data captured from a CMOS DSERN sensor. The results reveal the benefits of utilizing all available information in the raw sensor data and provide guidance on the optimal characterization method for different read noise regimes.

Single-pixel (SP) imaging and compressive sensing offer several advantages over traditional imaging. These include detection over a broader spectrum, robustness to light scattering, and lower data rates. When paired with low SWaP-C (Size, Weight, Power, and Cost), these benefits make single-pixel imaging a compelling alternative to conventional methods.

However, the widespread adoption of single-pixel imaging has been hindered by a significant trade-off between image quality and frame rate, while retaining a compelling SWaP-C.

We have developed a photonic integrated circuit (PIC) for Fourier-basis structured illumination that operates similarly to optical phased arrays, creating interferometric fringe patterns at up to 30 kHz. This refresh rate achieved through thermal-optic modulation offers a significant improvement over LCD and DMD based SP methods.

The scalability of PICs and advancements in laser hybrid integration are set to realize a commercially viable single-pixel imager with attractive SWaP-C, potentially transforming the imaging and sensing landscape.

We demonstrate the use of a free-running InGaAs/InP single-photon avalanche diode (SPAD) as the detector for a laser rangefinder (LRF). The bi-static LRF employs a laser with a wavelength of 1550 nm coupled to the transmit channel, and a receive channel fibre-coupled to the SPAD detector. The range measurement is based on the Time-Correlated Single-Photon Counting (TCSPC) technique with the Time-of-Flight (ToF) of the transmitted pulses being stored in a timing histogram. This has a time window of 134 µs, allowing a full detection range of up to ~ 20 km, with a distance resolution of ~ 30 cm. Due to its single-photon detection capabilities, the LRF can obtain long-range measurements in sub-second acquisition times using pulse energies as low as nano-Joules, requiring an average of 9 detected photons per target to achieve over 99% success rate. Here, we present the false alarm rate analysis of the SPAD detector based LRF. We also provide examples of range measurements of non-calibrated targets at distances of up to 18.9 km, using periodic pulses, as well as pseudo-random pulse patterns with a maximum average optical power of 9.8 mW and pulse energies between 3 nJ and 1.3 μJ.

We present a statistical model for the multiscale super-resolution of complex 3D single-photon LiDAR scenes while providing uncertainty measures about the depth and reflectivity parameters. We then propose a generalization of this model by unrolling its iterations into a new deep learning architecture which requires a reduced number of trainable parameters, and provides rich information about the estimates including uncertainty measures. The proposed algorithms will be demonstrated on two specific applications: micro-scanning with a 32 × 32 time-of-flight detector array, and sensor fusion for high-resolution kilometer-range 3D imaging. Results show that the proposed algorithms significantly enhance the image quality.

This paper presents a statistical approach for real-time 3D image reconstruction and target detection in challenging environments. The initial step analyzes photon events to detect targets, hence selecting informative pixels for data compression. The second contribution is a reconstruction algorithm that leverages data statistics and multiscale information to produce refined depth and reflectivity images, accompanied by uncertainty maps. The algorithms are implemented to run in parallel on GPUs, facilitating the real-time processing of moving scenes with over 50 depth frames-per-second for 192×128 pixel images. Results on real underwater data show the benefit of these algorithms when compared to existing ones.

Human monitoring using mmWave radar has recently become an area of significant research. The properties of radars makes them uniquely suited to imaging in adverse regimes such as through atmospheric obscurance or optically opaque media, such as housing materials. The privacy preserving nature of their data also allows radars to perform area monitoring and surveillance rolls with a reduced public impact. However, direct inference of human pose from radar data is challenging due to the relatively low transverse resolution of radar data. In this work we present a Convolutional Neural Network (CNN) capable of converting data from a commercially available Frequency Modulated Continuous Wave (FMCW) radar into human interpretable pose information. We employ a novel experimental configuration in which we combine a marker free motion capture suit with a single line sensing radar in an elevated position. We experimentally verify the ability of our system to reconstruct human pose and report average errors below 3 cm.

Detection in the short-wave infrared (SWIR) offers advantages like reduced solar noise and improved atmospheric transmission. Avalanche photodiodes (APDs) are ideal for low-light detection due to internal gain. While silicon (Si) APDs have low noise, they can't effectively detect SWIR light. Germanium (Ge) is good for SWIR detection but suffers from high noise. Ge-on-Si structure offers benefits like SWIR operation and efficient multiplication. This study showcases room temperature operation of a linear-mode pseudo-planar Ge-on-Si APD with high responsivity, gain, and low noise at 1550 nm. Moreover, a 10-pixel linear array exhibits uniform performance, promising for SWIR detection for potential LIDAR application.

Short-wave infrared (SWIR) image sensors play an important role in various defence and security applications, including low-light-level imaging, laser detection and range finding. Pixelated detectors made of indium gallium arsenide (InGaAs) have shown excellent performance for SWIR imaging, however, the cost remains the major barrier for adaptation.

In this talk we will present a low-cost, next-generation SWIR sensor. We have developed an alternative technology which relies on lead sulfide quantum dot (PbS QD) photon absorber, which is monolithically and directly deposited on a CMOS readout chip via solution-based process. The uncooled sensor achieves the QE of >20% at 1550 nm with the dark current of 50 nA/cm2. Additionally, the sensitivity range can be extended further to 2.1 um. The technology is suitable for wafer-level manufacturing, thus driving the sensor cost more than an order of magnitude lower. We will present a comprehensive characterization of QD-based imager and demonstrate real-life use cases.