Search

In the realm of autonomous driving, robust perception under out-of-distribution conditions is paramount for the safe deployment of vehicles. Challenges such as adverse weather, sensor malfunctions, and environmental unpredictability can severely impact the performance of autonomous systems. The 2024 RoboDrive Challenge was crafted to propel the development of driving perception technologies that can withstand and adapt to these real-world variabilities. Focusing on four pivotal tasks -- BEV detection, map segmentation, semantic occupancy prediction, and multi-view depth estimation -- the competition laid down a gauntlet to innovate and enhance system resilience against typical and atypical disturbances. This year's challenge consisted of five distinct tracks and attracted 140 registered teams from 93 institutes across 11 countries, resulting in nearly one thousand submissions evaluated through our servers. The competition culminated in 15 top-performing solutions, which introduced a range of innovative approaches including advanced data augmentation, multi-sensor fusion, self-supervised learning for error correction, and new algorithmic strategies to enhance sensor robustness. These contributions significantly advanced the state of the art, particularly in handling sensor inconsistencies and environmental variability. Participants, through collaborative efforts, pushed the boundaries of current technologies, showcasing their potential in real-world scenarios. Extensive evaluations and analyses provided insights into the effectiveness of these solutions, highlighting key trends and successful strategies for improving the resilience of driving perception systems. This challenge has set a new benchmark in the field, providing a rich repository of techniques expected to guide future research in this field., Comment: ICRA 2024; 32 pages, 24 figures, 5 tables; Code at https://robodrive-24.github.io/

The task of anomaly detection is to separate anomalous data from normal data in the dataset. Models such as deep convolutional autoencoder (CAE) network and deep supporting vector data description (SVDD) model have been universally employed and have demonstrated significant success in detecting anomalies. However, the over-reconstruction ability of CAE network for anomalous data can easily lead to high false negative rate in detecting anomalous data. On the other hand, the deep SVDD model has the drawback of feature collapse, which leads to a decrease of detection accuracy for anomalies. To address these problems, we propose the Improved AutoEncoder with LSTM module and Kullback-Leibler divergence (IAE-LSTM-KL) model in this paper. An LSTM network is added after the encoder to memorize feature representations of normal data. In the meanwhile, the phenomenon of feature collapse can also be mitigated by penalizing the featured input to SVDD module via KL divergence. The efficacy of the IAE-LSTM-KL model is validated through experiments on both synthetic and real-world datasets. Experimental results show that IAE-LSTM-KL model yields higher detection accuracy for anomalies. In addition, it is also found that the IAE-LSTM-KL model demonstrates enhanced robustness to contaminated outliers in the dataset. All code may be found at https://github.com/crazyn2/IAE-LSTM-KL_codes

By now, most outlier-detection algorithms struggle to accurately detect both point anomalies and cluster anomalies simultaneously. Furthermore, a few K-nearest-neighbor-based anomaly-detection methods exhibit excellent performance on many datasets, but their sensitivity to the value of K is a critical issue that needs to be addressed. To address these challenges, we propose a novel robust anomaly detection method, called Entropy Density Ratio Outlier Detection (EDROD). This method incorporates the probability density of each sample as the global feature, and the local entropy around each sample as the local feature, to obtain a comprehensive indicator of abnormality for each sample, which is called Entropy Density Ratio (EDR) for short in this paper. By comparing several competing anomaly detection methods on both synthetic and real-world datasets, it is found that the EDROD method can detect both point anomalies and cluster anomalies simultaneously with accurate performance. In addition, it is also found that the EDROD method exhibits strong robustness to the number of selected neighboring samples, the dimension of samples in the dataset, and the size of the dataset. Therefore, the proposed EDROD method can be applied to a variety of real-world datasets to detect anomalies with accurate and robust performances.

Quantum dot photonic crystal membrane lasers were fabricated and the large signal modulation characteristics were studied. We find that the modulation characteristics of quantum dot lasers can be significantly improved using cavities with large spontaneous emission coupling factor. Our experiments show, and simulations confirm, that the modulation rate is limited by the rate of carrier capture into the dots to around 30GHz in our present system.

We present a basic building block of a quantum network consisting of a quantum dot coupled to a source cavity, which in turn is coupled to a target cavity via a waveguide. The single photon emission from the high-Q/V source cavity is characterized by a twelve-fold spontaneous emission (SE) rate enhancement that results in a SE coupling efficiency near 0.98 into the source cavity mode. Single photons are efficiently transferred into the target cavity through the waveguide, with a source/target field intensity ratio of 0.12 (up to 0.49 observed in other structures without coupled quantum dots). This system shows great promise as a building block of future on-chip quantum information processing systems., Comment: 9 pages, 6 figures

We observe large spontaneous emission rate modification of individual InAs Quantum Dots (QDs) in 2D a photonic crystal with a modified, high-Q single defect cavity. Compared to QDs in bulk semiconductor, QDs that are resonant with the cavity show an emission rate increase by up to a factor of 8. In contrast, off-resonant QDs indicate up to five-fold rate quenching as the local density of optical states (LDOS) is diminished in the photonic crystal. In both cases we demonstrate photon antibunching, showing that the structure represents an on-demand single photon source with pulse duration from 210 ps to 8 ns. We explain the suppression of QD emission rate using Finite Difference Time Domain (FDTD) simulations and find good agreement with experiment., Comment: 11 pages, 6 figures

We have demonstrated efficient production of triggered single photons by coupling a single semiconductor quantum dot to a three-dimensionally confined optical mode in a micropost microcavity. The efficiency of emitting single photons into a single-mode travelling wave is approximately 38%, which is nearly two orders of magnitude higher than for a quantum dot in bulk semiconductor material. At the same time, the probability of having more than one photon in a given pulse is reduced by a factor of seven as compared to light with Poissonian photon statistics.