Labeling custom indoor point clouds through 2D semantic image segmentation. - In: 2022 Sixth IEEE International Conference on Robotic Computing, (2022), S. 261-264
For effective Computer Vision (CV) applications, one of the difficult challenges service robots have to face concerns with complete scene understanding. Therefore, various strategies are employed for point-level segregation of the 3D scene, such as semantic segmentation. Currently Deep Learning (DL) based algorithms are popular in this domain. However, they require precisely labeled ground truth data. Generating this data is a lengthy and expensive procedure, resulting in a limited variety of available data. On the contrary, the 2D image domain offers labeled data in abundance. Therefore, this study explores how we can achieve accurate labels for the 3D domain by utilizing semantic segmentation on 2D images and projecting the estimated labels to the 3D space via the depth channel. The labeled data may then be used for vision related tasks such as robot navigation or localization.
https://doi.org/10.1109/IRC55401.2022.00050
Emulation of electromagnetic plane waves for 3D antenna pattern estimation. - In: 2022 IEEE 96th Vehicular Technology Conference:(VTC 2022-Fall), (2022), insges. 6 S.
With the fast development of wireless devices, over-the-air (OTA) testing is becoming the preferred method among developers and manufacturers of wireless equipment. The ability to recreate a scenario under controllable and repeatable conditions keeps the method under constant development, providing new features that increase the realism during the tests. A recent proof of that is the integration of 3D wave field synthesis (3DWFS) to OTA testing, which becomes a significant step to accurately emulate wireless scenarios within a controlled environment.In this context, this contribution improves the OTA system calibration for 3DWFS; efficiently increasing the emulation quality of electromagnetic plane waves impinging from any angular position within an anechoic chamber. In fact, this enhancement implicitly delivers a new method for accurate estimation of the antenna radiation pattern in 3D. This is not only a highly demanded application among antenna manufacturers but in this case also proves the validity of the results and consolidates the integration of 3DWFS to OTA testing.
https://doi.org/10.1109/VTC2022-Fall57202.2022.10013008
Enhancement of vision-based 3D reconstruction systems using radar for smart farming. - In: 2022 IEEE International Workshop on Metrology for Agriculture and Forestry, (2022), S. 155-159
Digital field recordings are central to most precision agriculture systems since they can replicate the physical environment and thus monitor the state of an entire field or individual plants. Using different sensors, such as cameras and radar, data can be collected from various domains. Through the combination of radio wave propagation and visible light phenomena, it is possible to enhance, e.g., the optical condition of a fruit with internal parameters such as the water content. This paper proposes a method to correct sensor errors to perform data fusion. As an example, we observe a watermelon with camera and radar sensors and present a system architecture for the visualization of both sensors. For this purpose, we constructed a handheld platform on which both sensors are mounted. In our report, the radar is analyzed in terms of systematic and stochastic errors to formulate an angle-dependent mapping function for error correction. It is successfully shown that camera and radar data are correctly assigned with a watermelon used as a target object, demonstrated by a 3D reconstruction. The proposed system shows promising results for sensor overlay, but radar data remain challenging to interpret.
https://doi.org/10.1109/MetroAgriFor55389.2022.9964699
Deep learning aided interpolation of spatio-temporal nonstationary data. - In: 30th European Signal Processing Conference (EUSIPCO 2022), (2022), S. 2221-2225
Despite the growing interest in many fields, spatio-temporal (ST) interpolation remains challenging. Given ST nonstationary data distributed sparsely and irregularly over space, our objective is to obtain an equidistant representation of the region of interest (ROI). For this reason, an equidistant grid is defined within the ROI, where the available time series data are arranged, and the time series of the unobserved points are interpolated. Aiming to maintain the interpretability of the whole process while offering flexibility and fast execution, this work presents a ST interpolation frame-work which combines a statistical technique with deep learning. Our framework is generic and not confined to a specific application, which also provides the prediction confidence. To evaluate its validity, this framework is applied to ultrasound nondestructive testing (UT) data as an example. After the training with synthetic UT data sets, our framework is shown to yield accurate predictions when applied to measured UT data.
https://ieeexplore.ieee.org/document/9909600
Estimating multi-modal dense multipath components using auto-encoders. - In: 30th European Signal Processing Conference (EUSIPCO 2022), (2022), S. 1716-1720
We present a maximum-likelihood estimation algorithm for radio channel measurements exhibiting a mixture of independent Dense Multipath Components. The novelty of our approach is in the algorithms initialization using a deep learning architecture. Currently, available approaches can only deal with scenarios where a single mode is present. However, in measurements, two or more modes are often observed. This much more challenging multi-modal setting bears two important questions: How many modes are there, and how can we estimate those? To this end, we propose a Neural Net-architecture that can reliably estimate the number of modes present in the data and also provide an initial assessment of their shape. These predictions are used to initialize for gradient- and model-based optimization algorithm to further refine the estimates. We demonstrate numerically how the presented architecture performs on measurement data and analytically study its influence on the estimation of specular paths in a setting where the single-modal approach fails.
https://ieeexplore.ieee.org/document/9909796
Time-domain analysis of ultra-wideband scattering properties of fruits. - In: 2022 19th European Radar Conference, (2022), S. 77-80
In the present paper we evaluate scattering properties of fruits measured with a short-range Ultra-Wideband radar. This is part of our investigation how effectively such a radar can be used to infer information such as fruit biomass or ripeness in an agricultural environment. The covered frequency band spans from 1.4 to 5.6 GHz. We analyze measured impulse responses of a watermelon, a grapefruit, and an apple with respect to a dependency on the distance between radar and fruit and the observation angle i.e., rotation of the fruit. Measurements are performed under laboratory conditions, however, we analyze the data considering a pre-harvest analysis on a field. It becomes apparent that an analysis of the dispersed dominant reflection of the peel is most promising. Due to the natural growth and hence anisotropy of the fruits, we conclude to average over multiple monostatic observation angles to reduce the natural variations of e.g. the scattered power.
https://doi.org/10.23919/EuRAD54643.2022.9924720
Enable SDRs for real-time MIMO channel sounding featuring parallel coherent Rx channels. - In: 2022 IEEE 95th Vehicular Technology Conference (VTC2022-Spring), (2022), insges. 5 S.
A parallel receiver architecture for multiple input multiple output (MIMO) channel sounding application is presented with a software-defined radio (SDR)-based field-programmable gate array (FPGA) implementation. The receiver covers phase coherent reception via shared local oscillator (LO) and reference clock, a timing scheme synchronous to the antenna switching at the transmitter, and an integrated automatic gain control (AGC) in all receive channels. It is built with SDRs (NI USRP-2955, X310 series with TwinRx daughterboards). The use of these off-the-shelf hardware components reduces the costs of the sounding system. The FPGA implementation together with the system parameters of the chosen hardware allows a minimum AGC update interval of approx. 44.38 μs. Our setup demonstrates the applicability of state-of-the-art SDRs as a sounding system for continuous acquisition of the time variant, space, and frequency selective radio propagation channel.
https://doi.org/10.1109/VTC2022-Spring54318.2022.9860841
From 3D point cloud data to ray-tracing multi-band simulations in industrial scenario. - In: 2022 IEEE 95th Vehicular Technology Conference (VTC2022-Spring), (2022), insges. 5 S.
In this paper, we present the ray tracing (RT) simulation in the 3D model of one highly dense clutter industrial hall, which is scanned by laser scanner and reconstructed based on accurate point cloud. The whole processing chain from the scanning of the physical environment to running the simulation is presented in detail. To validate the simulation results, the synthetic channel characteristics and large-scale parameters, including delay spread (DS), angular spread (AS) and path loss (PL), are compared with those obtained from channel sounding measurement in both LOS and NLOS cases, at 6.75 GHz, 30 GHz and 60 GHz. The simulation results show that some scatters are significant in all bands and may be well identified and tracked. This indicates that our target to generate a deterministic channel model or a hybrid channel model at multi-band for industrial scenario may be possible.
https://doi.org/10.1109/VTC2022-Spring54318.2022.9861002
Reliable deep learning based localization with CSI fingerprints and multiple base stations. - In: 2022 IEEE International Conference on Communications, (2022), S. 3214-3219
Deep learning (DL) methods have been recently proposed for user equipment (UE) localization in wireless communication networks, based on the channel state information (CSI) between a UE and multiple base stations (BSs) in the uplink. With the CSI from the available BSs, UE localization can be performed in different ways. On the one hand, a single neural network (NN) can be trained for the UE localization by considering the CSI from all the available BSs as one overall fingerprint of the user’s location. On the other hand, the CSI at each BS can be used to obtain an estimate of the UE’s position with a separate NN at each BS, and then the position estimates of all BSs are combined to obtain an overall estimate of the UE position. In this work, we show that UE localization with the latter approach can achieve a higher positioning accuracy. We propose to consider the uncertainty in the UE localization at each BS, such that overall UE’s position is determined by combining the position estimates of the different BSs based on the uncertainty at each BS. With this approach, a more reliable position estimate can be obtained in case of variations in the channel.
https://doi.org/10.1109/ICC45855.2022.9838535
LTCC patch antenna array for 5G mobile applications featuring embedded air cavities. - In: 2022 International Conference on Electronics Packaging (ICEP 2022), (2022), S. 139-140
A LTCC patch antenna array was designed and fabricated in Low Temperature Co-fired Ceramic (LTCC) multilayer technology using picosecond laser structuring for precise manufacturing of embedded air cavities and feeding structures, which are beneficial for the antenna RF performance. Moreover, free-standing feeding vias in the embedded air cavities were successfully implemented in LTCC technology that enable a very simple and low-loss feed at the antenna ports. The 2x2 dual-polarized patch antenna array with embedded air cavities and transitions was first characterized by reflection measurements. The measured center frequency of the antenna is about 29.5 GHz and a bandwidth of nearly 1 GHz was achieved.
https://doi.org/10.23919/ICEP55381.2022.9795485