Autonomous Vehicles In Urban Environments Research Articles

Towards comprehensive understanding of pedestrians for autonomous driving: Efficient multi-task-learning-based pedestrian detection, tracking and attribute recognition

Safe interaction with pedestrians is important for autonomous vehicles in urban environments. Comprehensive understanding of pedestrian behavior is a challenge for autonomous driving, as different pedestrian attributes have different characteristics and there is not yet a dataset providing all attribute labels. In this paper, we design an efficient multi-task model for pedestrian detection, tracking and attribute recognition to fully mine information in multiple tasks and multiple datasets to reduce redundant image processing computation. Multiple training datasets are utilized to take full advantage of the available partially-labeled data. The model consists of two stages. The first stage is to detect and track pedestrians, using only low-resolution images to save computational resources. Accurate pedestrian detection is achieved in the bird's-eye-view (BEV), while pedestrian tracking is implemented on the basis of 3D bounding boxes. The second stage is image-based multi-attribute detection. This stage uses the detection and tracking results and some pre-processed information from the first stage along with high-resolution images. Multi-attributes of the pedestrian are identified, such as pedestrian's pose, age, gender, and motion direction. With a carefully-designed multi-task system that shares necessary feature maps among multiple tasks and reuses computation results, we can solve pedestrian detection, tracking, and multi-attribute recognition problems in an efficient framework. This leads to more accurate decisions and more efficient computations. The innovation of the model is to accomplish a simultaneous detection of multiple pedestrian attributes for autonomous driving, while fully sharing the features in the model to avoid repeated computation, as well as utilizing multiple partially-labeled datasets for the pedestrian detection, tracking and attribute recognition model training.

Semantic Communities from Graph-Inspired Visual Representations of Cityscapes

The swift development of autonomous vehicles raises the necessity of semantically mapping the environment by producing distinguishable representations to recognise similar areas. To this end, in this article, we present an efficient technique to cut up a robot’s trajectory into semantically consistent communities based on graph-inspired descriptors. This allows an agent to localise itself in future tasks under different environmental circumstances in an urban area. The proposed semantic grouping technique utilizes the Leiden Community Detection Algorithm (LeCDA), which is a novel and efficient method of low computational complexity and exploits semantic and topometric information from the observed scenes. The presented experimentation was carried out on a novel dataset from the city of Xanthi, Greece (dubbed as Gryphonurban urban dataset), which was recorded by RGB-D, IMU and GNSS sensors mounted on a moving vehicle. Our results exhibit the formulation of a semantic map with visually coherent communities and the realisation of an effective localisation mechanism for autonomous vehicles in urban environments.

A scalable time-division-based emergency messages broadcast scheme for connected and autonomous vehicles in urban environment

Connected and Autonomous Vehicles (CAVs) technology promises to revolutionize the transportation sector by solving many safety and traffic efficiency challenges while significantly enhancing passengers' and drivers' comfort. Most of these applications are built upon multi-hop broadcast of information among vehicles, leading to an increased contention and thus higher collision probability due to the limited bandwidth available for vehicular communications. In this paper, we propose a novel Time-Division-based Broadcast (TDB) scheme to mitigate interferences originating from hidden nodes (i.e., vehicles) to meet the stringent safety applications requirements. Then, we use this broadcast scheme as a basis for designing a new efficient and scalable protocol for emergency messages dissemination in urban vehicular networks (UV-TDB: an Urban, Vehicular and Time-Division-based Broadcast protocol). The simulation experiments show that UV-TDB significantly improves the performance of multi-hop broadcast and outperforms two state of the art protocols in terms of the achieved reduction in broadcast overhead (up to 77.9%), packet collisions ratio (up to 93.5%), interference rate (up to 43.5%) and increase in message reception rate (up to 337.7%). The results highlight also that UV-TDB adapts better to varying levels of vehicle density, making it more scalable.

Performance Evaluation of GNSS Position Augmentation Methods for Autonomous Vehicles in Urban Environments.

Global Navigation Satellite Systems provide autonomous vehicles with precise position information through the process of position augmentation. This paper presents a series of performance tests aimed to compare the position accuracy of augmentation techniques such as classical Differential Global Navigation Satellite System, Real-time Kinematic and Real-time eXtended. The aim is to understand the limitations and choose the best position augmentation technique in order to obtain accurate, trustworthy position estimates of a vehicle in urban environments. The tests are performed in and around the German cities of Wuppertal and Duesseldorf, using a vehicle fitted with the navigation system POS-LV 220, developed by Applanix Corporation. In order to evaluate the real-time performance of position augmentation techniques in a highly challenging environment, a total of four test regions are selected. The four test regions are characterized mainly by uneven terrain with tall buildings around the University of Wuppertal, flat terrain with roads of varying width in the city centre of Wuppertal and Duesseldorf and flat terrain in a tunnel section located in the city of Wuppertal. The performances of the different position augmentation are compared using a Root Mean Square (RMS) error estimate obtained as an output from the Applanix system. Furthermore, a High-Definition map of the environment is used for the purpose of model validation, which justifies the use of RMS error estimate as an evaluation metric for the performance analysis tests. According to the performance tests carried out as per the conditions specified in this paper, the Real-time eXtended (RTX) position augmentation method enables to obtain a more robust position information of the vehicle than Real-time Kinematic (RTK) method, with a typical accuracy of a few centimeter in an urban environment.

Decision-Making Method of Autonomous Vehicles in Urban Environments Considering Traffic Laws

In order to improve the efficiency and safety of autonomous vehicles’ decision-making process in complex urban scenarios, a decision-making method for hierarchical processing of static traffic law information and dynamic traffic participant information is proposed in this paper. In the decision-making method, the candidate behavior set is constructed by extracting the element of traffic laws and fully considering the traffic laws constraints to ensure the legality and effectiveness of the decision-making algorithm. Besides, four evaluation indicators and two-level entry threshold are designed to select the optimal driving behavior with the consideration of driving efficiency, ride safety and macro path requirement. The advantage of proposed method is that it avoids the problem of poor adaptability to traffic laws and regulations as the existing methods usually make decisions under the condition of mixed traffic laws and traffic participant information. A complete driving task simulation and analysis, including six typical urban traffic scenarios, is given under Matlab environment. The results show that the proposed decision-making method is able to make reasonable and feasible decisions and highly consistent with the actual driver’s decision-making behavior in complex urban scenarios, which verifies the effectiveness of the proposed method.

Recent Advances in Sensor Integrity Monitoring Methods—A Review

In face of the high complexity of modern systems and the increasing reliance on technology, the strategies and methods to evaluate the integrity of systems and the information they provide are of great importance. The use of sensors to provide important information in various applications has become ubiquitous, which puts more pressure on the accuracy of sensor measurements. In the navigation field, in particular, the integrity of positioning information provided by sensors can be critical, as failures can lead to fatalities and catastrophic damages. Extensive research has been done into methods to assess and improve the integrity of positioning information provided by GNSS receivers, led by the high safety requirements in the aviation field. More recently, a shift into the research of integrity solutions suitable for the challenging positioning requirements of autonomous vehicles in urban environment has been a trend. There have been extensive advancements in developing integrity monitoring algorithms for GNSS and navigation in general. However, these methods have not been sufficiently extended into other types of sensors, and the development of integrity methods outside the navigation field have been scarce and scattered. This work aims to provide an overview of recent advances in the integrity monitoring field, including research outside the navigation domain. The goal is to give an introduction to the integrity monitoring concept to a broader audience, as these techniques have been highly specialized by GNSS experts and navigation related research, fostering multi-disciplinary approaches and creative use of existing methods in different areas and applications.

VTGNet: A Vision-Based Trajectory Generation Network for Autonomous Vehicles in Urban Environments

Traditional methods for autonomous driving are implemented with many building blocks from perception, planning and control, making them difficult to generalize to varied scenarios due to complex assumptions and interdependencies. Recently, the end-to-end driving method has emerged, which performs well and generalizes to new environments by directly learning from export-provided data. However, many existing methods on this topic neglect to check the confidence of the driving actions and the ability to recover from driving mistakes. In this paper, we develop an uncertainty-aware end-to-end trajectory generation method based on imitation learning. It can extract spatiotemporal features from the front-view camera images for scene understanding, and then generate collision-free trajectories several seconds into the future. The experimental results suggest that under various weather and lighting conditions, our network can reliably generate trajectories in different urban environments, such as turning at intersections and slowing down for collision avoidance. Furthermore, closed-loop driving tests suggest that the proposed method achieves better cross-scene/platform driving results than the state-of-the-art (SOTA) end-to-end control method, where our model can recover from off-center and off-orientation errors and capture 80% of dangerous cases with high uncertainty estimations.

Safety Assessment and Simulation of Autonomous Vehicle in Urban Environments

The Autonomous Vehicle (AV) industry aims to design strategic plans to ensure the safety of the developed systems before their mass deployment. Real-road testing is shown to be impractical for validating these systems as it requires many years if not decades of testing in different environmental conditions. Therefore, this method must be complemented with simulation for a feasible solution. The primary goal of this research is to develop advanced techniques in the safety validation area by using immersive simulation technologies. This study led to a simulation approach for safety evaluation of an AV shuttle, ise Auto, currently operating at the TalTech campus. First, we create a virtual environment based on the specified path on the university campus including all relevant features. The environment is created by using geospatial data which is collected by a drone, then they are converted to a 3D map applicable for the LGSVL simulator. Next, the developed shuttle 3D model is imported to the simulation environment and with the aim of the LGSVL, sensor data are provided for the Autoware perception algorithms, which is the control software of the shuttle. This system enables us to evaluate AV’s decision-making performance and safety in different situations.

Path Tracking Control for Urban Autonomous Driving

Abstract A path tracking controller for autonomous vehicles in urban environments is presented. Based on system inversion, the steering angle causing the vehicle to follow the path in absence of disturbances is calculated. Then, the lateral distance and the orientation error w.r.t. the path are compensated by a state feedback controller. Further, a decoupling of the velocity is considered in the system-inversion and the feedback controller. Therefore, ideally, the velocity does not influence path tracking and, hence, the requirements on velocity control are relaxed. To reduce the effort for parameter identification, the controller is intentionally based on a kinematic vehicle model requiring less parameters compared to an elaborated dynamic model. It is assumed that the effects of unconsidered system components, e.g., tire slip, are then compensated by the state-feedback controller. The approach is validated on a closed proving-ground in a simulated urban scenario. Herein, for driving velocities up to 14 m/s and curve radii of down to 10 m, an RMS tracking error for the lateral distance to the path of 7.2 cm was achieved. The control system will be used in TomTom’s autonomous car ‘Trillian’ that serves as a validation and research platform to evaluate high definition maps of road networks.

Time-Dependent Hybrid-State A⁎ and Optimal Control for Autonomous Vehicles in Arbitrary and Dynamic Environments

Abstract The development of driving functions for autonomous vehicles in urban environments is still a challenging task. In comparison with driving on motorways, a wide variety of moving road users, such as pedestrians or cyclists, but also the strongly varying and sometimes very narrow road layout pose special challenges. The ability to make fast decisions about exact maneuvers and to execute them by applying sophisticated control commands is one of the key requirements for autonomous vehicles in such situations. In this context we present an algorithmic concept of three correlated methods. Its basis is a novel technique for the automated generation of a free-space polygon, providing a generic representation of the currently drivable area. We then develop a time-dependent hybrid-state A⁎ algorithm as a model-based planner for the efficient and precise computation of possible driving maneuvers in arbitrary dynamic environments. While on the one hand its results can be used as a basis for making short-term decisions, we also show their applicability as an initial guess for a subsequent trajectory optimization in order to compute applicable control signals. Finally, we provide numerical results for a variety of simulated situations demonstrating the efficiency and robustness of the proposed methods.

Practical Modeling of GNSS for Autonomous Vehicles in Urban Environments.

Autonomous navigation technology is used in various applications, such as agricultural robots and autonomous vehicles. The key technology for autonomous navigation is ego-motion estimation, which uses various sensors. Wheel encoders and global navigation satellite systems (GNSSs) are widely used in localization for autonomous vehicles, and there are a few quantitative strategies for handling the information obtained through their sensors. In many cases, the modeling of uncertainty and sensor fusion depends on the experience of the researchers. In this study, we address the problem of quantitatively modeling uncertainty in the accumulated GNSS and in wheel encoder data accumulated in anonymous urban environments, collected using vehicles. We also address the problem of utilizing that data in ego-motion estimation. There are seven factors that determine the magnitude of the uncertainty of a GNSS sensor. Because it is impossible to measure each of these factors, in this study, the uncertainty of the GNSS sensor is expressed through three variables, and the exact uncertainty is calculated. Using the proposed method, the uncertainty of the sensor is quantitatively modeled and robust localization is performed in a real environment. The approach is validated through experiments in urban environments.

A Hybrid Controller for Vision-Based Navigation of Autonomous Vehicles in Urban Environments

This paper presents a new hybrid control approach for vision-based navigation applied to autonomous robotic automobiles in urban environments. It is composed by a visual servoing (VS) for road lane following (as deliberative control) and a dynamic window approach (DWA) for obstacle avoidance (as reactive control). Typically, VS applications do not change the velocities to stop the robot in dangerous situations or avoid obstacles while performing the navigation task. However, in several urban conditions, these are elements that must be dealt with to guarantee the safe movement of the car. As a solution for this problem, in this paper, a line-following VS controller will be used to perform road lane following tasks with obstacle avoidance, validating its control outputs in a new image-based dynamic window approach (IDWA). The final solution combines the benefits of both controllers (VS+IDWA) for optimal lane following and fast obstacle avoidance, taking into account the car kinematics and some dynamics constraints. Experiments in a challenging scenario with both simulated and real experimental car show the viability of the proposed methodology.