arXiv subjects

Arkady Zgonnikov

Publications and source records attributed to Arkady Zgonnikov.

52 records · Page 3Linked to original sources

arXiv2023

A cognitive process approach to modeling gap acceptance in overtaking

Driving automation holds significant potential for enhancing traffic safety. However, effectively handling interactions with human drivers in mixed traffic remains a challenging task. Several models exist that attempt to capture human behavior in traffic interactions, often focusing on gap acceptance. However, it is not clear how models of an individual driver's gap acceptance can be translated to dynamic human-AV interactions in the context of high-speed scenarios like overtaking. In this study, we address this issue by employing a cognitive process approach to describe the dynamic interactions by the oncoming vehicle during overtaking maneuvers. Our findings reveal that by incorporating an initial decision-making bias dependent on the initial velocity into existing drift-diffusion models, we can accurately describe the qualitative patterns of overtaking gap acceptance observed previously. Our results demonstrate the potential of the cognitive process approach in modeling human overtaking behavior when the oncoming vehicle is an AV. To this end, this study contributes to the development of effective strategies for ensuring safe and efficient overtaking interactions between human drivers and AVs.

q-bio.NC↗

arXiv2023

Smooth-Trajectron++: Augmenting the Trajectron++ behaviour prediction model with smooth attention

Understanding traffic participants' behaviour is crucial for predicting their future trajectories, aiding in developing safe and reliable planning systems for autonomous vehicles. Integrating cognitive processes and machine learning models has shown promise in other domains but is lacking in the trajectory forecasting of multiple traffic agents in large-scale autonomous driving datasets. This work investigates the state-of-the-art trajectory forecasting model Trajectron++ which we enhance by incorporating a smoothing term in its attention module. This attention mechanism mimics human attention inspired by cognitive science research indicating limits to attention switching. We evaluate the performance of the resulting Smooth-Trajectron++ model and compare it to the original model on various benchmarks, revealing the potential of incorporating insights from human cognition into trajectory prediction models.

cs.LG↗

arXiv2023

Uncovering variability in human driving behavior through automatic extraction of similar traffic scenes from large naturalistic datasets

Recently, multiple naturalistic traffic datasets of human-driven trajectories have been published (e.g., highD, NGSim, and pNEUMA). These datasets have been used in studies that investigate variability in human driving behavior, for example for scenario-based validation of autonomous vehicle (AV) behavior, modeling driver behavior, or validating driver models. Thus far, these studies focused on the variability on an operational level (e.g., velocity profiles during a lane change), not on a tactical level (i.e., to change lanes or not). Investigating the variability on both levels is necessary to develop driver models and AVs that include multiple tactical behaviors. To expose multi-level variability, the human responses to the same traffic scene could be investigated. However, no method exists to automatically extract similar scenes from datasets. Here, we present a four-step extraction method that uses the Hausdorff distance, a mathematical distance metric for sets. We performed a case study on the highD dataset that showed that the method is practically applicable. The human responses to the selected scenes exposed the variability on both the tactical and operational levels. With this new method, the variability in operational and tactical human behavior can be investigated, without the need for costly and time-consuming driving-simulator experiments.

cs.CV↗

arXiv2023

Benchmark for Models Predicting Human Behavior in Gap Acceptance Scenarios

Autonomous vehicles currently suffer from a time-inefficient driving style caused by uncertainty about human behavior in traffic interactions. Accurate and reliable prediction models enabling more efficient trajectory planning could make autonomous vehicles more assertive in such interactions. However, the evaluation of such models is commonly oversimplistic, ignoring the asymmetric importance of prediction errors and the heterogeneity of the datasets used for testing. We examine the potential of recasting interactions between vehicles as gap acceptance scenarios and evaluating models in this structured environment. To that end, we develop a framework aiming to facilitate the evaluation of any model, by any metric, and in any scenario. We then apply this framework to state-of-the-art prediction models, which all show themselves to be unreliable in the most safety-critical situations.

cs.RO↗

arXiv2022

Haptic Shared Control for Dissipating Phantom Traffic Jams

Traffic jams occurring on highways cause increased travel time as well as increased fuel consumption and collisions. Traffic jams without a clear cause, such as an on-ramp or an accident, are called phantom traffic jams and are said to make up 50% of all traffic jams. They are the result of an unstable traffic flow caused by human driving behavior. Automating the longitudinal vehicle motion of only 5% of all cars in the flow can dissipate phantom traffic jams. However, driving automation introduces safety issues when human drivers need to take over the control from the automation. We investigated whether phantom traffic jams can be dissolved using haptic shared control. This keeps humans in the loop and thus bypasses the problem of humans' limited capacity to take over control, while benefiting from most advantages of automation. In an experiment with 24 participants in a driving simulator, we tested the effect of haptic shared control on the dynamics of traffic flow, and compared it with manual control and full automation. We also investigated the effect of two control types on participants' behavior during simulated silent automation failures. Results show that haptic shared control can help dissipating phantom traffic jams better than fully manual control but worse than full automation. We also found that haptic shared control reduces the occurrence of unsafe situations caused by silent automation failures compared to full automation. Our results suggest that haptic shared control can dissipate phantom traffic jams while preventing safety risks associated with full automation.

cs.HC↗

arXiv2022

A human factors approach to validating driver models for interaction-aware automated vehicles

A major challenge for autonomous vehicles is interacting with other traffic participants safely and smoothly. A promising approach to handle such traffic interactions is equipping autonomous vehicles with interaction-aware controllers (IACs). These controllers predict how surrounding human drivers will respond to the autonomous vehicle's actions, based on a driver model. However, the predictive validity of driver models used in IACs is rarely validated, which can limit the interactive capabilities of IACs outside the simple simulated environments in which they are demonstrated. In this paper, we argue that besides evaluating the interactive capabilities of IACs, their underlying driver models should be validated on natural human driving behavior. We propose a workflow for this validation that includes scenario-based data extraction and a two-stage (tactical/operational) evaluation procedure based on human factors literature. We demonstrate this workflow in a case study on an inverse-reinforcement-learning-based driver model replicated from an existing IAC. This model only showed the correct tactical behavior in 40% of the predictions. The model's operational behavior was inconsistent with observed human behavior. The case study illustrates that a principled evaluation workflow is useful and needed. We believe that our workflow will support the development of appropriate driver models for future automated vehicles.

cs.RO↗

arXiv2022

Meaningful human control: actionable properties for AI system development

How can humans remain in control of artificial intelligence (AI)-based systems designed to perform tasks autonomously? Such systems are increasingly ubiquitous, creating benefits - but also undesirable situations where moral responsibility for their actions cannot be properly attributed to any particular person or group. The concept of meaningful human control has been proposed to address responsibility gaps and mitigate them by establishing conditions that enable a proper attribution of responsibility for humans; however, clear requirements for researchers, designers, and engineers are yet inexistent, making the development of AI-based systems that remain under meaningful human control challenging. In this paper, we address the gap between philosophical theory and engineering practice by identifying, through an iterative process of abductive thinking, four actionable properties for AI-based systems under meaningful human control, which we discuss making use of two applications scenarios: automated vehicles and AI-based hiring. First, a system in which humans and AI algorithms interact should have an explicitly defined domain of morally loaded situations within which the system ought to operate. Second, humans and AI agents within the system should have appropriate and mutually compatible representations. Third, responsibility attributed to a human should be commensurate with that human's ability and authority to control the system. Fourth, there should be explicit links between the actions of the AI agents and actions of humans who are aware of their moral responsibility. We argue that these four properties will support practically-minded professionals to take concrete steps toward designing and engineering for AI systems that facilitate meaningful human control.

cs.CY↗

arXiv2021

MORAL: Aligning AI with Human Norms through Multi-Objective Reinforced Active Learning

Inferring reward functions from demonstrations and pairwise preferences are auspicious approaches for aligning Reinforcement Learning (RL) agents with human intentions. However, state-of-the art methods typically focus on learning a single reward model, thus rendering it difficult to trade off different reward functions from multiple experts. We propose Multi-Objective Reinforced Active Learning (MORAL), a novel method for combining diverse demonstrations of social norms into a Pareto-optimal policy. Through maintaining a distribution over scalarization weights, our approach is able to interactively tune a deep RL agent towards a variety of preferences, while eliminating the need for computing multiple policies. We empirically demonstrate the effectiveness of MORAL in two scenarios, which model a delivery and an emergency task that require an agent to act in the presence of normative conflicts. Overall, we consider our research a step towards multi-objective RL with learned rewards, bridging the gap between current reward learning and machine ethics literature.

cs.LG↗

arXiv2020

Optimality and limitations of audio-visual integration for cognitive systems

Multimodal integration is an important process in perceptual decision-making. In humans, this process has often been shown to be statistically optimal, or near optimal: sensory information is combined in a fashion that minimises the average error in perceptual representation of stimuli. However, sometimes there are costs that come with the optimization, manifesting as illusory percepts. We review audio-visual facilitations and illusions that are products of multisensory integration, and the computational models that account for these phenomena. In particular, the same optimal computational model can lead to illusory percepts, and we suggest that more studies should be needed to detect and mitigate these illusions, as artefacts in artificial cognitive systems. We provide cautionary considerations when designing artificial cognitive systems with the view of avoiding such artefacts. Finally, we suggest avenues of research towards solutions to potential pitfalls in system design. We conclude that detailed understanding of multisensory integration and the mechanisms behind audio-visual illusions can benefit the design of artificial cognitive systems.

cs.AI↗

arXiv2018

Complexity of human response delay in intermittent control: The case of virtual stick balancing

Response delay is an inherent and essential part of human actions. In the context of human balance control, the response delay is traditionally modeled using the formalism of delay-differential equations, which adopts the approximation of fixed delay. However, experimental studies revealing substantial variability, adaptive anticipation, and non-stationary dynamics of response delay provide evidence against this approximation. In this paper, we call for development of principally new mathematical formalism describing human response delay. To support this, we present the experimental data from a simple virtual stick balancing task. Our results demonstrate that human response delay is a widely distributed random variable with complex properties, which can exhibit oscillatory and adaptive dynamics characterized by long-range correlations. Given this, we argue that the fixed-delay approximation ignores essential properties of human response, and conclude with possible directions for future developments of new mathematical notions describing human control.

q-bio.NC↗

arXiv2016

Noise-induced phase transition in the model of human virtual stick balancing

Humans face the task of balancing dynamic systems near an unstable equilibrium repeatedly throughout their lives. Much research has been aimed at understanding the mechanisms of intermittent control in the context of human balance control. The present paper deals with one of the recent developments in the theory of human intermittent control, namely, the double-well model of noise-driven control activation. We demonstrate that the double-well model can reproduce the whole range of experimentally observed distributions under different conditions. Moreover, we show that a slight change in the noise intensity parameter leads to a sudden shift of the action point distribution shape, that is, a phase transition is observed.

q-bio.NC↗

arXiv2015

Statistical Properties of Car Following: Theory and Driving Simulator Experiments

A fair simple car driving simulator was created based on the open source engine TORCS and used in car-following experiments aimed at studying the basic features of human behavior in car driving. Four subjects with different skill in driving real cars participated in these experiments. The subjects were instructed to drive a car without overtaking and losing sight of a lead car driven by computer at a fixed speed. Based on the collected data the distributions of the headway distance, the car velocity, acceleration, and jerk are constructed and compared with the available experimental data for the real traffic flow. A new model for the car-following is proposed to capture the found properties. As the main result, we draw a conclusion that human actions in car driving should be categorized as generalized intermittent control with noise-driven activation. Besides, we hypothesize that the car jerk together with the car acceleration are additional phase variables required for describing the dynamics of car motion governed by human drivers.

nlin.AO↗

arXiv2015

Bistable dynamics of control activation in human intermittent control

When facing a task of balancing a dynamic system near an unstable equilibrium, humans often adopt intermittent control strategy: instead of continuously controlling the system, they repeatedly switch the control on and off. Paradigmatic example of such a task is stick balancing. Despite the simplicity of the task itself, the complexity of human intermittent control dynamics in stick balancing still puzzles researchers in motor control. Here we attempt to model one of the key mechanisms of human intermittent control, control activation, using as an example the task of overdamped stick balancing. In so doing, we focus on the concept of noise-driven activation, a more general alternative to the conventional threshold-driven activation. We describe control activation as a random walk in an energy potential, which changes in response to the state of the controlled system. By way of numerical simulations, we show that the developed model captures the core properties of human control activation observed previously in the experiments on overdamped stick balancing. Our results demonstrate that the double-well potential model provides tractable mathematical description of human control activation at least in the considered task, and suggest that the adopted approach can potentially aid in understanding human intermittent control in more complex processes.

physics.bio-ph↗

arXiv2014

To react or not to react? Intrinsic stochasticity of human control in virtual stick balancing

Understanding how humans control unstable systems is central to many research problems, with applications ranging from quiet standing to aircraft landing. Increasingly much evidence appears in favor of event-driven control hypothesis: human operators only start actively controlling the system when the discrepancy between the current and desired system states becomes large enough. The event-driven models based on the concept of threshold can explain many features of the experimentally observed dynamics. However, much still remains unclear about the dynamics of human-controlled systems, which likely indicates that humans employ more intricate control mechanisms. The present paper argues that control activation in humans may be not threshold-driven, but instead intrinsically stochastic, noise-driven. Specifically, we suggest that control activation stems from stochastic interplay between the operator's need to keep the controlled system near the goal state on one hand and the tendency to postpone interrupting the system dynamics on the other hand. We propose a model capturing this interplay and show that it matches the experimental data on human balancing of virtual overdamped stick. Our results illuminate that the noise-driven activation mechanism plays a crucial role at least in the considered task, and, hypothetically, in a broad range of human-controlled processes.

physics.bio-ph↗

arXiv2014

Unstable Dynamics of Adaptation in Unknown Environment due to Novelty Seeking

Learning and adaptation play great role in emergent socio-economic phenomena. Complex dynamics has been previously found in the systems of multiple learning agents interacting via a simple game. Meanwhile, the single agent adaptation is considered trivially stable. We advocate the idea that adopting a more complex model of the individual behavior may result in a more diverse spectrum of macro-level behaviors. We develop an adaptation model based on the reinforcement learning framework extended by an additional processing channel. We scrutiny the dynamics of the single agent adapting to the unknown environment; the agent is biased by novelty seeking, the intrinsic inclination for exploration. We demonstrate that the behavior of the novelty-seeking agent may be inherently unstable. One of the surprising results is that under certain conditions the increase of the novelty-seeking level may cause the agent to switch from the non-rational to the strictly rational behavior. Our results give evidence to the hypothesis that the intrinsic motives of agents should be paid no less attention than the extrinsic ones in the models of complex socio-economic systems.

physics.soc-ph↗

arXiv2014

Extended phase space description of human-controlled systems dynamics

Humans are often incapable of precisely identifying and implementing the desired control strategy in controlling unstable dynamical systems. That is, the operator of a dynamical system treats the current control effort as acceptable even if it deviates slightly from the desired value, and starts correcting the actions only when the deviation has become evident. We argue that the standard Newtonian approach does not allow to model such behavior. Instead, the physical phase space of a controlled system should be extended with an independent phase variable characterizing the operator motivated actions. The proposed approach is illustrated via a simple non-Newtonian model capturing the operators fuzzy perception of their own actions. The properties of the model are investigated analytically and numerically; the results confirm that the extended phase space may aid in capturing the intricate dynamical properties of human-controlled systems.

nlin.AO↗