Exploration of the acceptability of different behaviors of an autonomous vehicle in so-called conflict situations

The acceptability of autonomous vehicles is an important issue for car manufacturers. The subject of this work attempts to address this issue in the context of conflict situations in urban environments. We present the results of a preliminary study that was conducted to investigate the effects of the driving mode and context on the acceptability of the behavior of an autonomous vehicle. We thus measured acceptability on 30 drivers confronted with three driving modes (defensive, aggressive, and transgressive) as well as different situations constituted by the most common urban intersections in France. We then formulated hypotheses on the effects that the driving mode, the context, or the socio-demographic characteristics of the passengers may have on their acceptability of the autonomous vehicle behavior. In the context of our study, the driving mode of the vehicle was the most influential parameter on the level of acceptability assessed by the participants. The type of intersection used did not result in a significant difference, nor did the socio-demographic characteristics studied. The results of these works allow an interesting first approach and lead us to guide our future works on the study of the parameters involved in the driving modes of autonomous vehicles.

Who goes first? A distributed simulator study of vehicle-pedestrian interaction

One of the current challenges of automation is to have highly automated vehicles (HAVs) that communicate effectively with pedestrians and react to changes in pedestrian behaviour, to promote more trustable HAVs. However, the details of how human drivers and pedestrians interact at unsignalised crossings remain poorly understood. We addressed some aspects of this challenge by replicating vehicle-pedestrian interactions in a safe and controlled virtual environment by connecting a high fidelity motion-based driving simulator to a CAVE-based pedestrian lab in which 64 participants (32 pairs of one driver and one pedestrian) interacted with each other under different scenarios. The controlled setting helped us study the causal role of kinematics and priority rules on interaction outcome and behaviour, something that is not possible in naturalistic studies. We also found that kinematic cues played a stronger role than psychological traits like sensation seeking and social value orientation in determining whether the pedestrian or driver passed first at unmarked crossings. One main contribution of this study is our experimental paradigm, which permitted repeated observation of crossing interactions by each driver-pedestrian participant pair, yielding behaviours which were qualitatively in line with observations from naturalistic studies.

Longitudinal traffic conflict analysis of autonomous and traditional vehicle platoons in field tests via surrogate safety measures

Autonomous vehicles (AVs) have been introduced into the traffic stream alongside traditional vehicles (TVs) with the expectation of improved transportation safety, efficiency, and reliability. The majority of AV safety research has been done through simulation. The results of such research on the safety performances of AVs are heavily influenced by the methodological framework, algorithms, and assumptions about AV driving characteristics in a simulated environment. There is a need for AV safety research based on real-world settings before any wide-scale deployment of this technology. This paper investigates the impact of the presence of SAE level 2 AVs in the traffic stream in reducing longitudinal traffic conflicts using Surrogate Safety Measures on a real-world open-source database of mixed traffic trajectories. The analysis is conducted for both AV-exclusive and mixed AV-TV platoons. Furthermore, we explore whether the presence of AVs decreases longitudinal traffic conflicts in two-vehicle platoons comprising AV and TV mixed leaders and followers. We find that an exclusive AV platoon behaves similarly to an exclusive TV platoon and produces similar longitudinal conflicts. However, mixed platoons with both AVs and TVs result in a higher number of longitudinal conflicts. Maintaining near-identical leader-follower conditions, we find that the number of conflicts in mixed platoons when an AV follows a TV is higher than when a TV follows an AV. The increase in conflict numbers in a TV-AV mixed platoon can be attributed to AV's longer response time lag. In summary, analyses conducted in this paper indicate that exclusive platoons and pairs of vehicles exhibit fewer longitudinal conflicts than mixed platoons and pairs.

Self-supervised recurrent depth estimation with attention mechanisms

Depth estimation has been an essential task for many computer vision applications, especially in autonomous driving, where safety is paramount. Depth can be estimated not only with traditional supervised learning but also via a self-supervised approach that relies on camera motion and does not require ground truth depth maps. Recently, major improvements have been introduced to make self-supervised depth prediction more precise. However, most existing approaches still focus on single-frame depth estimation, even in the self-supervised setting. Since most methods can operate with frame sequences, we believe that the quality of current models can be significantly improved with the help of information about previous frames. In this work, we study different ways of integrating recurrent blocks and attention mechanisms into a common self-supervised depth estimation pipeline. We propose a set of modifications that utilize temporal information from previous frames and provide new neural network architectures for monocular depth estimation in a self-supervised manner. Our experiments on the KITTI dataset show that proposed modifications can be an effective tool for exploiting temporal information in a depth prediction pipeline.

State Laws for Autonomous Vehicle Safety, Equity, and Insurance

This article reviews existing state laws related to autonomous vehicle (AV) safety, equity, and automobile insurance. Thirty states were identified with relevant legislation. Of these, most states had one or two relevant laws in place. Many of these laws were related to safety and insurance requirements. Data are needed to evaluate the effectiveness of these laws in order to guide further policy development.

Safety analysis of freeway on-ramp merging with the presence of autonomous vehicles

Freeway on-ramp merging areas are high-risk areas for motor vehicle crashes and conflicts due to the variety of driving styles, the difference in mainline and ramp traffic states, and factors related to roadway design and traffic control. The emerging Autonomous Vehicle (AV) technologies are expected to bring substantial improvements in ramp merging operations in general, including the possibility to reduce traffic conflicts and crashes by partially or fully eliminating the critical factors related to the human drivers. In order to investigate the potential safety impacts of AVs at on-ramp merging, this study first proposes a novel conflict index in theory as a specific indicator for ramp merging safety. Then, a merging conflict model is introduced to estimate the index value in various cases by considering the interactions between the mainline and ramp vehicles. In order to account for real-world uncertainties and variations in various crash-contributing factors, the proposed approach incorporates Monte-Carlo method and probabilistic distributions calibrated on the empirical freeway data. The developed approach is later applied in a case study with incremental AV market penetration rates to investigate AV safety impacts at on-ramp merging. The results show clear benefits of AVs in reducing the frequency and severity of the critical merging events. In addition, a sensitivity analysis on essential model parameters shows that the merging safety of AVs is closely related to their gap acceptance policy and the proper functioning of the driving systems, providing further insights into the future development of AVs.

High Precision Outdoor and Indoor Reference State Estimation for Testing Autonomous Vehicles

A current trend in automotive research is autonomous driving. For the proper testing and validation of automated driving functions a reference vehicle state is required. Global Navigation Satellite Systems (GNSS) are useful in the automation of the vehicles because of their practicality and accuracy. However, there are situations where the satellite signal is absent or unusable. This research work presents a methodology that addresses those situations, thus largely reducing the dependency of Inertial Navigation Systems (INSs) on the SatNav. The proposed methodology includes (1) a standstill recognition based on machine learning, (2) a detailed mathematical description of the horizontation of inertial measurements, (3) sensor fusion by means of statistical filtering, (4) an outlier detection for correction data, (5) a drift detector, and (6) a novel LiDAR-based Positioning Method (LbPM) for indoor navigation. The robustness and accuracy of the methodology are validated with a state-of-the-art INS with Real-Time Kinematic (RTK) correction data. The results obtained show a great improvement in the accuracy of vehicle state estimation under adverse driving conditions, such as when the correction data is corrupted, when there are extended periods with no correction data and in the case of drifting. The proposed LbPM method achieves an accuracy closely resembling that of a system with RTK.

Online supervised attention-based recurrent depth estimation from monocular video

Autonomous driving highly depends on depth information for safe driving. Recently, major improvements have been taken towards improving both supervised and self-supervised methods for depth reconstruction. However, most of the current approaches focus on single frame depth estimation, where quality limit is hard to beat due to limitations of supervised learning of deep neural networks in general. One of the way to improve quality of existing methods is to utilize temporal information from frame sequences. In this paper, we study intelligent ways of integrating recurrent block in common supervised depth estimation pipeline. We propose a novel method, which takes advantage of the convolutional gated recurrent unit (convGRU) and convolutional long short-term memory (convLSTM). We compare use of convGRU and convLSTM blocks and determine the best model for real-time depth estimation task. We carefully study training strategy and provide new deep neural networks architectures for the task of depth estimation from monocular video using information from past frames based on attention mechanism. We demonstrate the efficiency of exploiting temporal information by comparing our best recurrent method with existing image-based and video-based solutions for monocular depth reconstruction.

The Road to Zero: A Vision for Achieving Zero Roadway Deaths by 2050

Imagine that, in 2050, not a single person in the United States dies in a traffic crash. This is the scenario described in this article, in which RAND researchers set forth a vision and strategy for achieving zero roadway deaths by 2050. The authors propose that a combination of three approaches can realize this scenario. The first is doubling down on programs and policies that have already been shown to be effective, including laws and enforcement, changes to roadway infrastructure designed to reduce traffic conflicts, reductions in speeds where crashes are likely, improvements to emergency response and trauma care, and more safety education and outreach. The second is accelerating advanced technology, beginning with advanced driver assistance systems (many of which are already in the market) and progressing up to fully automated vehicles. The third is prioritizing safety, which includes both (1) embracing a new safety culture that will lead Americans to think differently about our individual and collective choices and (2) widespread adoption of the "Safe System" approach, a paradigm shift in addressing the causes and prevention of roadway deaths and injuries. The authors conclude with a list of actions that key stakeholders-including professional engineering and planning organizations, public-sector organizations, safety advocates, vehicle manufacturers, technology developers, public health, emergency medical and trauma care organizations, and law enforcement and judicial system representatives-can take to bring about the changes needed to achieve zero roadway deaths by 2050.