Search

Robust environment perception for autonomous vehicles is a tremendous challenge, which makes a diverse sensor set with e.g. camera, lidar and radar crucial. In the process of understanding the recorded sensor data, 3D semantic segmentation plays an important role. Therefore, this work presents a pyramid-based deep fusion architecture for lidar and camera to improve 3D semantic segmentation of traffic scenes. Individual sensor backbones extract feature maps of camera images and lidar point clouds. A novel Pyramid Fusion Backbone fuses these feature maps at different scales and combines the multimodal features in a feature pyramid to compute valuable multimodal, multi-scale features. The Pyramid Fusion Head aggregates these pyramid features and further refines them in a late fusion step, incorporating the final features of the sensor backbones. The approach is evaluated on two challenging outdoor datasets and different fusion strategies and setups are investigated. It outperforms recent range view based lidar approaches as well as all so far proposed fusion strategies and architectures., Comment: conditionally accepted at IEEE IV 2022, 7 pages, 4 figures, 5 tables

Understanding and interpreting a 3d environment is a key challenge for autonomous vehicles. Semantic segmentation of 3d point clouds combines 3d information with semantics and thereby provides a valuable contribution to this task. In many real-world applications, point clouds are generated by lidar sensors in a consecutive fashion. Working with a time series instead of single and independent frames enables the exploitation of temporal information. We therefore propose a recurrent segmentation architecture (RNN), which takes a single range image frame as input and exploits recursively aggregated temporal information. An alignment strategy, which we call Temporal Memory Alignment, uses ego motion to temporally align the memory between consecutive frames in feature space. A Residual Network and ConvGRU are investigated for the memory update. We demonstrate the benefits of the presented approach on two large-scale datasets and compare it to several stateof-the-art methods. Our approach ranks first on the SemanticKITTI multiple scan benchmark and achieves state-of-the-art performance on the single scan benchmark. In addition, the evaluation shows that the exploitation of temporal information significantly improves segmentation results compared to a single frame approach.

Reflective imaging systems form an important part of photonic devices such as spectrometers, telescopes, augmented and virtual reality headsets or lithography platforms. Reflective optics provide unparalleled spectral performance and can be used to reduce overall volume and weight. So far, most reflective designs have focused on two or three reflections, while four-reflection freeform designs can deliver a higher light throughput (faster F-number) as well as a larger field-of-view (FOV). However, advanced optical design strategies for four-reflection freeform systems have been rarely reported in literature. This is due to the increased complexity in solution space but also the fact that additional mirrors hinder a cost-effective realization (manufacture, alignment, etc.). Recently, we have proposed a novel design method to directly calculate the freeform surface coefficients while merely knowing the mirror positions and tilts. Consequently, this method allows laymen with basic optical design knowledge to calculate 'first time right' freeform imaging systems in a matter of minutes. This contrasts with most common freeform design processes, which requires considerable experience, intuition or guesswork. Firstly, we demonstrate the effectiveness of the proposed method for a four-mirror high-throughput telescope with 250mm-focal-length, F/2.5 and a wide rectangular FOV of 8.5{\deg} x 25.5{\deg}. In a subsequent step, we propose an effective three-mirror but four-reflection imaging system, which consists of two freeform mirrors and one double-reflection spherical mirror. Compared with common three-mirror and three-reflection imagers, our novel multi-reflection system shows unprecedented possibilities for an economic implementation while drastically reducing the overall volume., Comment: 18 pages, 8 figures, submitted to journal

For more than 150 years, scientists have advanced aberration theory to describe, analyse, and eliminate imperfections that disturb the imaging quality of optical components and systems. Simultaneously, they have developed optical design methods for and manufacturing techniques of imaging systems with ever-increasing complexity and performance up to the point where they are now including optical elements that are unrestricted in their surface shape. These so-called optical freeform elements offer degrees of freedom that can greatly extend the functionalities and further boost the specifications of state-of-the-art imaging systems. However, the drastically increased number of surface coefficients of these freeform surfaces poses severe challenges for the optical design process, such that the deployment of freeform optics remained limited until today. In this paper, we present a deterministic direct optical design method for freeform imaging systems based on differential equations derived from Fermat's principle and solved using power series. The method allows calculating the optical surface coefficients that ensure minimal image blurring for each individual order of aberrations. We demonstrate the systematic, deterministic, scalable and holistic character of our method with catoptric and catadioptric design examples. As such we introduce a disruptive methodology to design optical imaging systems 'first time right', we eliminate the 'trial and error' approach in present-day optical design, and we pave the way to a fast-track uptake of freeform elements to create the next-generation high-end optics., Comment: 15 pages, 5 figures, submitted to journal

Quadratic detection in linear mesoscopic transport systems produces cross terms that can be viewed as interference signals reflecting statistical properties of charge carriers. In electronic systems these cross term interferences arise from exchange effects due to Pauli principle. Here we demonstrate fermionic Hanbury Brown and Twiss (HBT) exchange phenomena due to indistinguishability of charge carriers in a diffusive graphene system. These exchange effects are verified using current-current cross correlations in combination with regular shot noise (autocorrelation) experiments at microwave frequencies. Our results can be modeled using semiclassical analysis for a square-shaped metallic diffusive conductor, including contributions from contact transparency. The experimentally determined HBT exchange factor values lie between the calculated ones for coherent and hot electron transport., Comment: 25 pages, 7 figures

We have investigated current-current correlations in a cross-shaped conductor made of graphene ribbons. We measured auto and cross correlations and compared them with the theoretical predictions for ideal diffusive conductors. Our data deviate from these predictions and agreement can be obtained only by adding contributions from occupation-number noise in the central region connecting the arms of the cross. Furthermore, we have determined Hanbury -- Brown and Twiss (HBT) exchange correlations in this system. Contrary to expectations for a cross-shaped diffusive system, we find finite HBT exchange effects due to the occupation-number noise at the crossing. The strength of these HBT exchange correlations is found to vary with gate voltage, and very a distinct HBT effect with large fluctuations is observed near the Dirac point., Comment: 5 + 3 pages, 5 + 1 figures

The presence of strong disorder in graphene nanoribbons yields low-mobility diffusive transport at high charge densities, whereas a transport gap occurs at low densities. Here, we investigate the longitudinal and transverse magnetoresistance of a narrow (60 nm) nanoribbon in a six-terminal Hall bar geometry. At B= 11 T, quantum Hall plateaux appear at $\sigma_{xy}=\pm2e^2/h$, $\pm6e^2/h$ and $\pm10e^2/h$, for which the Landau level spacing is larger than the Landau level broadening. Interestingly, the transport gap does not disappear in the quantum Hall regime, when the zero-energy Landau level is present at the charge neutrality point, implying that it cannot originate from a lateral confinement gap. At high charge densities, the longitudinal and Hall resistance exhibit reproducible fluctuations, which are most pronounced at the transition regions between Hall plateaux. Bias-dependent measurements strongly indicate that these fluctuations can be attributed to phase coherent scattering in the disordered ribbon., Comment: experimental paper; 4 pages, 4 figures

In graphene, which is an atomic layer of crystalline carbon, two of the distinguishing properties of the material are the charge carriers two-dimensional and relativistic character. The first experimental evidence of the two-dimensional nature of graphene came from the observation of a sequence of plateaus in measurements of its transport properties in the presence of an applied magnetic field. These are signatures of the so-called integer quantum Hall effect. However, as a consequence of the relativistic character of the charge carriers, the integer quantum Hall effect observed in graphene is qualitatively different from its semiconductor analogue. As a third distinguishing feature of graphene, it has been conjectured that interactions and correlations should be important in this material, but surprisingly, evidence of collective behaviour in graphene is lacking. In particular, the quintessential collective quantum behaviour in two dimensions, the fractional quantum Hall effect (FQHE), has so far resisted observation in graphene despite intense efforts and theoretical predictions of its existence. Here we report the observation of the FQHE in graphene. Our observations are made possible by using suspended graphene devices probed by two-terminal charge transport measurements. This allows us to isolate the sample from substrate-induced perturbations that usually obscure the effects of interactions in this system and to avoid effects of finite geometry. At low carrier density, we find a field-induced transition to an insulator that competes with the FQHE, allowing its observation only in the highest quality samples. We believe that these results will open the door to the physics of FQHE and other collective behaviour in graphene., Comment: 14 pages, 3 figures