How Do Autonomous Agents and Drivers Behave? An Analysis of Micro Accidents in Autonomous Driving Videos

In this paper, the urgent evacuation behavior of passengers at aircraft accident is formulated as a simulation model which consists of two models such as a two-dimensional grid cell model and an autonomous and multiple agents' system model. The former model represents the layout of cabin interior and the allocation of passengers. On the other hand, the latter shows the behavior of evacuation of passengers from inside of a cabin to outside of an aircraft using emergency equipments such as an emergency exit and escaping slide. This autonomous agent is able to obtain the necessary information for urgent evacuation within one's field of view and decide to select the escaping route by means of obtained information concerned with escaping path to the nearest emergency exit. By the comparable study between the results of the proposed simulation and the accident analysis reports on “Garuda Indonesia Airways Accident on 1996” and “China Airline Accident on 2007 ,” it is verified that the proposed simulation system enable one to estimate the situation of urgent evacuation in the aircraft accident.

At present, it is expected that pedestrian agent simulation will be applied to not only accident analysis, but also spatial design; ASPF (Agent-based Simulator of Pedestrian Flows) has already been developed as a simulator for such purposes. However, in the present version ASPFver.3 a pedestrian agent merely walks straight ahead and simply avoids other agents, and it had been impossible to analyze crowd flows on a large-scale space with a complicated shape, a function is required that enables an agent to walk along a chain of visible target ‘waypoints’ to each destination, as well as a function the agent keeps the direction to the target. The study introduces newly a target maintaining (Helmsman) function, a concept of waypoint, and update mechanism of targets, and develops the simulator ASP-Fver4.0 that models an autonomous pedestrian agent on ArtiSoc(KKMAS). The performances tests of these additional functions of ASPF ver 4.0 are shown. Especially, to successfully model pedestrians’ shop-around behavior in a Patio-style shopping mall at Asunal Kanayama, Nagoya, ASPF ver 4.1 has been also developed by introducing an optimization function of routes by Dijkstra method, and implemented several parameters based on data for survey of the pedestrians’ behaviors in this mall. Through the test of four simulation cases; (1) weekday case, (2) weekday double case, (3) holiday case, and (4) at time of a music event in holiday case, the performance of ASPFver4.1 was also verified. Due to a series of these version-ups, we can conclude that ASPF is now available for analyzing crowd flows and density in space with complicated shapes.

Evacuation flow analysis for aircraft accident with panic passenger by autonomous agent and multi-agent simulation

As information machines become increasingly autonomous, not only partial and predetermined aspects of contracting are delegated to them, but also the very deliberations pertaining to whether to enter a contract, with whom, and with what content. After considering the different aspects of autonomy, I argue that the use of autonomous contracting agents involves the delegation of cognitive tasks. Consequently, such agents should be considered the real authors of the contracts they conclude. Not only do they exercise discretion in contracting, but they also have those cognitive states that are relevant to the contracts they form (beliefs, goals and intentions). Consequently, the rules that apply to human agents should in principle also apply to autonomous digital agents. This does not presuppose that digital agents should be granted legal personality, understood as the ability to bear rights and duties, since the normative effects of their activity will fall upon their users. Finally, I consider some of the legal and social issues that may result from the widespread delegation of contracting to digital agents. INTRODUCTION Contracting takes place today in a new socio-technological space, the so-called infosphere, based on digital information and information machines. In such a context, several contract-related tasks are either accomplished with the support of machines, or are completely delegated to them. As information machines become increasingly autonomous, not only are partial and predetermined aspects of contracting delegated to them, but also the very deliberations pertaining to whether to enter a contract, with whom, and with what content. I shall address some emerging dimensions of contracting in the ICT-based infosphere, focusing on artificially intelligent contracting parties. First, I shall address the embeddedness of contractual activities in active, responsive, and connected informational environments. I will argue that in such environments a vast expansion of contractual interactions is likely to happen, since contracting enables distributed autonomous collaboration between humans and artificial entities, and between such entities themselves. Contractual relations will be entered into by both virtual artefacts (e.g., online bots) and physical artefacts (e.g., autonomous cars), as interactions with physical artefacts take place through their interconnected digital interfaces. I will then focus on how contracting may take place with and between Autonomous Contracting Agents – ‘ACA’ s. I will characterise the idea of autonomy as including three aspects: independence of action, high-level cognitive skills, and adaptive/teleologic architecture.

In this paper we introduce and experimentally evaluate a new suboptimal decision-making design to be used by autonomous agents acting on behalf of a user in repeated tasks, whenever the agent's autonomy level is continuously controlled by the user. This mode of operation is common and can be found whenever user's perception of the agent's competence is affected by the nature of the outcomes resulting from the agent's decisions rather than the optimality of the decisions made, e.g., in spam filtering, CV filtering, poker agents, and robotic vacuum cleaners as well as in newly arriving systems such as autonomous cars. Our proposed design relies on choosing the action that offers the best tradeoff between decision optimality and the influence over future allowed autonomy, where the latter is predicted using standard machine learning techniques. The design is found to be highly effective compared to following the theoreticoptimal decision rule, over various measures, through extensive experimentation with a virtual investment agent, making virtual investments on behalf of 679 subjects using Amazon Mechanical Turk.

Autonomous driving of automobiles is a hot research topic in recent years. The autonomous driving tractor also has been studied in the agricultural field as well as an autonomous driving automobile. On the other hand, tractor accidents frequently occur on the farm. Tractor accident can be a major obstacle for autonomous driving tractor because farm operation by tractor would be stopped if the accident occurs. Therefore, accident analysis of tractor is very important for the development of autonomous driving tractor. In this study, numerical analysis of tractor accident was conducted using commercial driving simulator CarSim®. Typical two accident cases, that is falling accident and overturning accident, were considered in the numerical experiments. Numerical results obtained in the study shows that the driving simulator is capable of reproducing above accident cases. Therefore, the driving simulator can be a strong platform for the research of accident analysis and autonomous driving.

The recent years several companies have invested a lot of money on autonomous cars. This is a part of the future, where autonomous cars (i.e. smart-□cars) will transport us to different places without drivers in a nearly optimum way. It should be noted that we are not ready yet, although companies promote their autonomous cars with 80% reliable performance. Thus in this paper, we present a smart car model based on intelligent agents for reducing accident using a more safe approach. The approach is based on existing traffic rules and by training the smart cars on human drivers'safe behavior.

Road safety is a global concern impacting millions of individuals. Governments, industry leaders, and researchers are actively collaborating to enhance same through various techniques, and technologies. Among them Advanced Driver Assistance Systems (ADAS), Autonomous Driving Systems (ADSs), and Connected Vehicle Technologies (CVT) are promising candidates with the potential to mitigate accident risks and improve overall road safety. Autonomous driving is still emerging and needs to address potential challenges which includes maneuvering over the curves. This paper describes the process of design and implementation of an autonomous driving agent based on deep reinforcement learning techniques to boost safety within the realm of autonomous driving. The agent is developed and tested in CARLA simulator, aiming to address the challenges associated with reliability of ADSs. The results show a clear improvement over the average deviation from the centre of the road and distance covered. Maneuvering over the curves is effectively handled by the proposed system and keeps the car to stay on the road rather than going off-road.

This paper focuses on the collaboration between human drivers and intelligent vehicles. We propose a collaboration mechanism grounded on the concept of distributed cognition. With distributed cognition, intelligence does not lie just in the single entity but also in the interaction with the other cognitive components in a system. We apply this idea to vehicle intelligence, proposing a system distributed into two cognitive entities—the human and the autonomous agent—that together contribute to drive the vehicle. This account of vehicle intelligence differs from the mainstream research effort on highly autonomous cars. The proposed mechanism follows one of the paradigm derived from distributed cognition, the rider-horse metaphor: just like the rider communicates their intention to the horse through the reins, the human influences the agent using the pedals and the steering wheel. We use a driving simulator to demonstrate the collaboration in action, showing how the human can communicate and interact with the agent in various ways with safe outcomes.

Intrinsically, driving is a Markov Decision Process which suits well the reinforcement learning paradigm. In this paper, we propose a novel agent which learns to drive a vehicle without any human assistance. We use the concept of reinforcement learning and evolutionary strategies to train our agent in a 2D simulation environment. Our model's architecture goes beyond the World Model's by introducing difference images in the auto encoder. This novel involvement of difference images in the auto-encoder gives better representation of the latent space with respect to the motion of vehicle and helps an autonomous agent to learn more efficiently how to drive a vehicle. Results show that our method requires fewer (96% less) total agents, (87.5% less) agents per generations, (70% less) generations and (90% less) rollouts than the original architecture while achieving the same accuracy of the original.

The impact of the temperament model on the behavior of an autonomous driver

Studying driver behaviors has become a major concern for the transportation community, businesses, and the public. Thus, based on the simulation, we proposed an adaptive driving model in the car-following driving behavior and based on the normative behavior of the driver during decision-making and anticipation, whose intention is to ensure the objectives of imitation of ordinary human behavior and road safety. The presented model is based on a software agent paradigm to model a human driver and the Fuzzy Logic Theory to reflect the driver agent’s reasoning. To validate our model, we used the dataset from the program of the US Federal Highway Administration. In this context, we notice an excellent homogeneity in the deviation of the adopted trajectory of the autonomous driver agent from the adopted trajectories by the human drivers. Moreover, the advantage of our model is that it works with different velocities.

Emotions Embodied in the SVC of an Autonomous Driver System

The autonomous car technology promises to replace human drivers with safer driving systems. But although autonomous cars can become safer than human drivers this is a long process that is going to be refined over time. Before these vehicles are deployed on urban roads a minimum safety level must be assured. Since the autonomous car technology is still under development there is no standard methodology to evaluate such systems. It is important to completely understand the technology that is being developed to design efficient means to evaluate it. In this paper we assume safety-critical systems reliability as a safety measure. We model an autonomous road vehicle as an intelligent agent and we approach its evaluation from an artificial intelligence perspective. Our focus is the evaluation of perception and decision-making systems and also to propose a systematic method to evaluate their integration in the vehicle. We identify critical aspects of the data dependency from the artificial intelligence state of the art models and we also propose procedures to evaluate them.

Navigating roundabouts is a complex driving scenario for both manual and autonomous vehicles. This paper proposes an approach based on the use of the Q-learning algorithm to train an autonomous vehicle agent to learn how to appropriately navigate roundabouts. The proposed learning algorithm is implemented using the CARLA simulation environment. Several simulations are performed to train the algorithm in two scenarios: navigating a roundabout with and without surrounding traffic. The results illustrate that the Q-learning-algorithm-based vehicle agent is able to learn smooth and efficient driving to perform maneuvers within roundabouts.