Advanced eNose-Driven Pedestrian Tracking Pipeline for Intelligent Car Driver Assisting System: Preliminary Results

From a biological point of view, alcohol human attentional impairment occurs before reaching a Blood Alcohol Content (BAC index) of 0.08% (0.05% under the Italian legislation), thus generating a significant impact on driving safety if the drinker subject is driving a car. Car drivers must keep a safe driving dynamic, having an unaltered physiological status while processing the surrounding information coming from the driving scenario (e.g., traffic signs, other vehicles and pedestrians). Specifically, the identification and tracking of pedestrians in the driving scene is a widely investigated problem in the scientific community. The authors propose a full, deep pipeline for the identification, monitoring and tracking of the salient pedestrians, combined with an intelligent electronic alcohol sensing system to properly assess the physiological status of the driver. More in detail, the authors propose an intelligent sensing system that makes a common air quality sensor selective to alcohol. A downstream Deep 1D Temporal Residual Convolutional Neural Network architecture will be able to learn specific embedded alcohol-dynamic features in the collected sensing data coming from the GHT25S air-quality sensor of STMicroelectronics. A parallel deep attention-augmented architecture identifies and tracks the salient pedestrians in the driving scenario. A risk assessment system evaluates the sobriety of the driver in case of the presence of salient pedestrians in the driving scene. The collected preliminary results confirmed the effectiveness of the proposed approach.

$$360^{\circ }$$ real-time and power-efficient 3D DAMOT for autonomous driving applications

Autonomous Driving (AD) promises an efficient, comfortable and safe driving experience. Nevertheless, fatalities involving vehicles equipped with Automated Driving Systems (ADSs) are on the rise, especially those related to the perception module of the vehicle. This paper presents a real-time and power-efficient 3D Multi-Object Detection and Tracking (DAMOT) method proposed for Intelligent Vehicles (IV) applications, allowing the vehicle to track $$360^{\circ }$$ 360 ∘ surrounding objects as a preliminary stage to perform trajectory forecasting to prevent collisions and anticipate the ego-vehicle to future traffic scenarios. First, we present our DAMOT pipeline based on Fast Encoders for object detection and a combination of a 3D Kalman Filter and Hungarian Algorithm, used for state estimation and data association respectively. We extend our previous work ellaborating a preliminary version of sensor fusion based DAMOT, merging the extracted features by a Convolutional Neural Network (CNN) using camera information for long-term re-identification and obstacles retrieved by the 3D object detector. Both pipelines exploit the concepts of lightweight Linux containers using the Docker approach to provide the system with isolation, flexibility and portability, and standard communication in robotics using the Robot Operating System (ROS). Second, both pipelines are validated using the recently proposed KITTI-3DMOT evaluation tool that demonstrates the full strength of 3D localization and tracking of a MOT system. Finally, the most efficient architecture is validated in some interesting traffic scenarios implemented in the CARLA (Car Learning to Act) open-source driving simulator and in our real-world autonomous electric car using the NVIDIA AGX Xavier, an AI embedded system for autonomous machines, studying its performance in a controlled but realistic urban environment with real-time execution (results).

New human-computer interaction regulations for ship safe driving based on multi-mode data interconnection

Based on the analysis of the multi-mode data of ship mechatronics and the new human-computer interaction regulations for safety driving, a new safety driving regulation based on multi-mode data is put forward. The new regulations for ship safe driving use mechanical and electrical data to form small-world data interconnection. Artificial intelligence and human-computer interaction operation information are used to integrate and communicate, and human-computer interaction data are incorporated to standardize driving behavior to integrate historical driving data, and finally, the standardized automatic self-driving is formed. The new human-computer interaction regulations formed by the safe driving system make it possible to solve and optimize the ship safe driving mode.

Research on artificial intelligence safety prediction and intervention model based on ship driving habits

Based on the analysis of the causes of ship accidents, the development prospect and development direction of ship intelligent safe driving, the artificial intelligence safety prediction and intervention model is put forward. This model solves the problem of ship intelligent safety prediction by using intelligent analysis technology and network technology, and promotes the development of ship intelligence and ship safety navigation technology. Additionally, it expands the channels of obtaining information, connects the ship's mechanical and electrical equipment, collects, stores and analyzes the data reasonably, and constructs the intelligent analysis and processing platform of ship small-world data processing to implement intelligent intervention. What is impressive is that it makes ship navigation safer, more economical, more reasonable and optimized, and accelerates the development of ship artificial intelligence safe navigation.

A Survey of Localization Methods for Autonomous Vehicles in Highway Scenarios

In the context of autonomous vehicles on highways, one of the first and most important tasks is to localize the vehicle on the road. For this purpose, the vehicle needs to be able to take into account the information from several sensors and fuse them with data coming from road maps. The localization problem on highways can be distilled into three main components. The first one consists of inferring on which road the vehicle is currently traveling. Indeed, Global Navigation Satellite Systems are not precise enough to deduce this information by themselves, and thus a filtering step is needed. The second component consists of estimating the vehicle’s position in its lane. Finally, the third and last one aims at assessing on which lane the vehicle is currently driving. These two last components are mandatory for safe driving as actions such as overtaking a vehicle require precise information about the current localization of the vehicle. In this survey, we introduce a taxonomy of the localization methods for autonomous vehicles in highway scenarios. We present each main component of the localization process, and discuss the advantages and drawbacks of the associated state-of-the-art methods.

IMPROVING AUTOMOBILE INSURANCE CLAIMS FREQUENCY PREDICTION WITH TELEMATICS CAR DRIVING DATA

Novel navigation applications provide a driving behavior score for each finished trip to promote safe driving, which is mainly based on experts’ domain knowledge. In this paper, with automobile insurance claims data and associated telematics car driving data, we propose a supervised driving risk scoring neural network model. This one-dimensional convolutional neural network takes time series of individual car driving trips as input and returns a risk score in the unit range of (0,1). By incorporating credibility average risk score of each driver, the classical Poisson generalized linear model for automobile insurance claims frequency prediction can be improved significantly. Hence, compared with non-telematics-based insurers, telematics-based insurers can discover more heterogeneity in their portfolio and attract safer drivers with premiums discounts.

A Driving Simulation to Analysis and Quantitative Comparison of Driving Behavior of Guide Signs at Intersections

Guide signs are an important source for drivers to obtain road information. However, the evaluation methods for the effectiveness of guide signs are not unified. The quantitative model for evaluating guide signs needs to be constructed to unify the current system of guide signs. This study aims to take the commonly used guide signs in China as the research object to explore the evaluation method of guide signs at intersections. Eight kinds of guide signs were designed and made based on the common layout (layout 1 and layout 2) and the amount of information on signs (3–6). Thirty-four drivers were recruited to organize a driving simulation based on the visual cognitive tasks. Drivers’ legibility time and driver behavior were obtained by using the driving simulator and E-Prime program. A comprehensive quantitative evaluation model of guide signs was established based on the factor analysis method and grey correlation analysis method from the perspective of safe driving. The results show that there is no significant difference in the SD of speed and the SD of acceleration under the influence of various guide signs. The average vehicle speed and acceleration decrease, and the lateral offset distance of the vehicle increases with the amount of information on guide signs increasing. The quantitative evaluation results of guide signs show that the visual security decreases with the increase of the amount of information on guide signs. And layout 2 has better performance than layout 1 when the amount of information on guide signs is the same. This study not only explores the change rule of driving behavior under the influence of guide signs, but also provides a reference for the selection of guide signs.

Safe Driving Of Autonomous Vehicles Through Improved Deep Reinforcement Learning

In this thesis, we propose an environment perception framework for autonomous driving using deep reinforcement learning (DRL) that exhibits learning in autonomous vehicles under complex interactions with the environment, without being explicitly trained on driving datasets. Unlike existing techniques, our proposed technique takes the learning loss into account under deterministic as well as stochastic policy gradient. We apply DRL to object detection and safe navigation while enhancing a self-driving vehicle’s ability to discern meaningful information from surrounding data. For efficient environmental perception and object detection, various Q-learning based methods have been proposed in the literature. Unlike other works, this thesis proposes a collaborative deterministic as well as stochastic policy gradient based on DRL. Our technique is a combination of variational autoencoder (VAE), deep deterministic policy gradient (DDPG), and soft actor-critic (SAC) that adequately trains a self-driving vehicle. In this work, we focus on uninterrupted and reasonably safe autonomous driving without colliding with an obstacle or steering off the track. We propose a collaborative framework that utilizes best features of VAE, DDPG, and SAC and models autonomous driving as partly stochastic and partly deterministic policy gradient problem in continuous action space, and continuous state space. To ensure that the vehicle traverses the road over a considerable period of time, we employ a reward-penalty based system where a higher negative penalty is associated with an unfavourable action and a comparatively lower positive reward is awarded for favourable actions. We also examine the variations in policy loss, value loss, reward function, and cumulative reward for ‘VAE+DDPG’ and ‘VAE+SAC’ over the learning process.

Safe Driving Of Autonomous Vehicles Through Improved Deep Reinforcement Learning

In this thesis, we propose an environment perception framework for autonomous driving using deep reinforcement learning (DRL) that exhibits learning in autonomous vehicles under complex interactions with the environment, without being explicitly trained on driving datasets. Unlike existing techniques, our proposed technique takes the learning loss into account under deterministic as well as stochastic policy gradient. We apply DRL to object detection and safe navigation while enhancing a self-driving vehicle’s ability to discern meaningful information from surrounding data. For efficient environmental perception and object detection, various Q-learning based methods have been proposed in the literature. Unlike other works, this thesis proposes a collaborative deterministic as well as stochastic policy gradient based on DRL. Our technique is a combination of variational autoencoder (VAE), deep deterministic policy gradient (DDPG), and soft actor-critic (SAC) that adequately trains a self-driving vehicle. In this work, we focus on uninterrupted and reasonably safe autonomous driving without colliding with an obstacle or steering off the track. We propose a collaborative framework that utilizes best features of VAE, DDPG, and SAC and models autonomous driving as partly stochastic and partly deterministic policy gradient problem in continuous action space, and continuous state space. To ensure that the vehicle traverses the road over a considerable period of time, we employ a reward-penalty based system where a higher negative penalty is associated with an unfavourable action and a comparatively lower positive reward is awarded for favourable actions. We also examine the variations in policy loss, value loss, reward function, and cumulative reward for ‘VAE+DDPG’ and ‘VAE+SAC’ over the learning process.

Lateral and Longitudinal Driving Behavior Prediction Based on Improved Deep Belief Network

Accurately predicting driving behavior can help to avoid potential improper maneuvers of human drivers, thus guaranteeing safe driving for intelligent vehicles. In this paper, we propose a novel deep belief network (DBN), called MSR-DBN, by integrating a multi-target sigmoid regression (MSR) layer with DBN to predict the front wheel angle and speed of the ego vehicle. Precisely, the MSR-DBN consists of two sub-networks: one is for the front wheel angle, and the other one is for speed. This MSR-DBN model allows ones to optimize lateral and longitudinal behavior predictions through a systematic testing method. In addition, we consider the historical states of the ego vehicle and surrounding vehicles and the driver’s operations as inputs to predict driving behaviors in a real-world environment. Comparison of the prediction results of MSR-DBN with a general DBN model, back propagation (BP) neural network, support vector regression (SVR), and radical basis function (RBF) neural network, demonstrates that the proposed MSR-DBN outperforms the others in terms of accuracy and robustness.