YOLOPoint Joint Keypoint and Object Detection

Intelligent vehicles of the future must be capable of understanding and navigating safely through their surroundings. Camera-based vehicle systems can use keypoints as well as objects as low- and high-level landmarks for GNSS-independent SLAM and visual odometry. To this end we propose YOLOPoint, a convolutional neural network model that simultaneously detects keypoints and objects in an image by combining YOLOv5 and SuperPoint to create a single forward-pass network that is both real-time capable and accurate. By using a shared backbone and a light-weight network structure, YOLOPoint is able to perform competitively on both the HPatches and KITTI benchmarks.

Low Latency Instance Segmentation by Continuous Clustering for Rotating LiDAR Sensors

Low-latency instance segmentation of LiDAR point clouds is crucial in real-world applications because it serves as an initial and frequently-used building block in a robot's perception pipeline, where every task adds further delay. Particularly in dynamic environments, this total delay can result in significant positional offsets of dynamic objects, as seen in highway scenarios. To address this issue, we employ continuous clustering of obstacle points in order to obtain an instance-segmented point cloud. Unlike most existing approaches, which use a full revolution of the LiDAR sensor, we process the data stream in a continuous and seamless fashion. More specifically, each column of a range image is processed as soon it is available. Obstacle points are clustered to existing instances in real-time and it is checked at a high-frequency which instances are completed and are ready to be published. An additional advantage is that no problematic discontinuities between the points of the start and the end of a scan are observed. In this work we describe the two-layered data structure and the corresponding algorithm for continuous clustering, which is able to cluster the incoming data in real time. We explain the importance of a large perceptive field of view. Furthermore, we describe and evaluate important architectural design choices, which could be relevant to design an architecture for deep learning based low-latency instance segmentation. We are publishing the source code at https://github.com/UniBwTAS/continuous_clustering.

The GOOSE Dataset for Perception in Unstructured Environments

The potential for deploying autonomous systems can be significantly increased by improving the perception and interpretation of the environment. However, the development of deep learning-based techniques for autonomous systems in unstructured outdoor environments poses challenges due to limited data availability for training and testing. To address this gap, we present the German Outdoor and Offroad Dataset (GOOSE), a comprehensive dataset specifically designed for unstructured outdoor environments. The GOOSE dataset incorporates 10 000 labeled pairs of images and point clouds, which are utilized to train a range of state-of-the-art segmentation models on both image and point cloud data. We open source the dataset, along with an ontology for unstructured terrain, as well as dataset standards and guidelines. This initiative aims to establish a common framework, enabling the seamless inclusion of existing datasets and a fast way to enhance the perception capabilities of various robots operating in unstructured environments. The dataset, pre-trained models for offroad perception, and additional documentation can be found at https://goose-dataset.de/.

TAS-NIR: A VIS+NIR Dataset for Fine-grained Semantic Segmentation in Unstructured Outdoor Environments

Vegetation Indices based on paired images of the visible color spectrum (VIS) and near infrared spectrum (NIR) have been widely used in remote sensing applications. These vegetation indices are extended for their application in autonomous driving in unstructured outdoor environments. In this domain we can combine traditional vegetation indices like the Normalized Difference Vegetation Index (NDVI) and Enhanced Vegetation Index (EVI) with Convolutional Neural Networks (CNNs) pre-trained on available VIS datasets. By laying a focus on learning calibrated CNN outputs, we can provide an approach to fuse known hand-crafted image features with CNN predictions for different domains as well. The method is evaluated on a VIS+NIR dataset of semantically annotated images in unstructured outdoor environments. The dataset is available at mucar3.de/iros2022-ppniv-tas-nir.

A Fine-Grained Dataset and its Efficient Semantic Segmentation for Unstructured Driving Scenarios

Research in autonomous driving for unstructured environments suffers from a lack of semantically labeled datasets compared to its urban counterpart. Urban and unstructured outdoor environments are challenging due to the varying lighting and weather conditions during a day and across seasons. In this paper, we introduce TAS500, a novel semantic segmentation dataset for autonomous driving in unstructured environments. TAS500 offers fine-grained vegetation and terrain classes to learn drivable surfaces and natural obstacles in outdoor scenes effectively. We evaluate the performance of modern semantic segmentation models with an additional focus on their efficiency. Our experiments demonstrate the advantages of fine-grained semantic classes to improve the overall prediction accuracy, especially along the class boundaries. The dataset and pretrained model are available at mucar3.de/icpr2020-tas500.

Combining Deep Learning and Model-Based Methods for Robust Real-Time Semantic Landmark Detection

Compared to abstract features, significant objects, so-called landmarks, are a more natural means for vehicle localization and navigation, especially in challenging unstructured environments. The major challenge is to recognize landmarks in various lighting conditions and changing environment (growing vegetation) while only having few training samples available. We propose a new method which leverages Deep Learning as well as model-based methods to overcome the need of a large data set. Using RGB images and light detection and ranging (LiDAR) point clouds, our approach combines state-of-the-art classification results of Convolutional Neural Networks (CNN), with robust model-based methods by taking prior knowledge of previous time steps into account. Evaluations on a challenging real-wold scenario, with trees and bushes as landmarks, show promising results over pure learning-based state-of-the-art 3D detectors, while being significant faster.

The Greedy Dirichlet Process Filter - An Online Clustering Multi-Target Tracker

Reliable collision avoidance is one of the main requirements for autonomous driving. Hence, it is important to correctly estimate the states of an unknown number of static and dynamic objects in real-time. Here, data association is a major challenge for every multi-target tracker. We propose a novel multi-target tracker called Greedy Dirichlet Process Filter (GDPF) based on the non-parametric Bayesian model called Dirichlet Processes and the fast posterior computation algorithm Sequential Updating and Greedy Search (SUGS). By adding a temporal dependence we get a real-time capable tracking framework without the need of a previous clustering or data association step. Real-world tests show that GDPF outperforms other multi-target tracker in terms of accuracy and stability.