Abstract:The interactions of motorised vehicles with pedestrians have always been a concern in traffic safety. The major threat to pedestrians comes from the high level of interactions imposed in uncontrolled traffic environments, where road users have to compete over the right of way. In the absence of traffic management and control systems in such traffic environments, road users have to negotiate the right of way while avoiding conflict. Furthermore, the high level of movement freedom and agility of pedestrians, as one of the interactive parties, can lead to exposing unpredictable behaviour on the road. Traffic interactions in uncontrolled mixed traffic environments will become more challenging by fully/partially automated driving systems' deployment, where the intentions and decisions of interacting agents must be predicted/detected to avoid conflict and improve traffic safety and efficiency. This study aims to formulate a game-theoretic approach to model pedestrian interactions with passenger cars and light vehicles (two-wheel and three-wheel vehicles) in uncontrolled traffic settings. The proposed models employ the most influencing factors in the road user's decision and choice of strategy to predict their movements and conflict resolution strategies in traffic interactions. The models are applied to two data sets of video recordings collected in a shared space in Hamburg and a mid-block crossing area in Surat, India, including the interactions of pedestrians with passenger cars and light vehicles, respectively. The models are calibrated using the identified conflicts between users and their conflict resolution strategies in the data sets. The proposed models indicate satisfactory performances considering the stochastic behaviour of road users – particularly in the mid-block crossing area in India – and have the potential to be used as a behavioural model for automated driving systems.

Modelling motorized and non-motorized vehicle conflicts using multiagent inverse reinforcement learning approach

A cooperative mobility model for multiple autonomous vehicles

Multiagent modeling of pedestrian-vehicle conflicts using Adversarial Inverse Reinforcement Learning

Microscopic Traffic Simulation by Cooperative Multi-agent Deep Reinforcement Learning

Modelling two-dimensional driving behaviours at unsignalised intersection using multi-agent imitation learning

Modeling pedestrian behavior in pedestrian-vehicle near misses: A continuous Gaussian Process Inverse Reinforcement Learning (GP-IRL) approach

A game-theoretic approach for modelling pedestrian–vehicle conflict resolutions in uncontrolled traffic environments

Multi-agent reinforcement learning for traffic congestion on one-way multi-lane highways

Multi-agent Based Modeling and Simulation of Microscopic Traffic in Virtual Reality System.

Multiagent reinforcement learning for autonomous driving in traffic zones with unsignalized intersections

Brainstorming Multi-Agent Reinforcement Learning for Multi-Vehicles Games

Game-Theoretic Modeling of Multi-Vehicle Interactions at Uncontrolled Intersections

Objective-aware Traffic Simulation via Inverse Reinforcement Learning

Towards a systematic computational framework for modeling multi-agent decision-making at micro level for smart vehicles in a smart world

Conflict Resolution Model of Automated Vehicles Based on Multi-Vehicle Cooperative Optimization at Intersections

Adversarial and Reactive Traffic Agents for Realistic Driving Simulation

A Cooperative Multiagent System for Traffic Signal Control Using Game Theory and Reinforcement Learning

Decentralized Non-communicating Multiagent Collision Avoidance with Deep Reinforcement Learning

Modeling Interactions of Autonomous Vehicles and Pedestrians with Deep Multi-Agent Reinforcement Learning for Collision Avoidance

A Joint Prediction Method of Multi-Agent to Reduce Collision Rate

Cooperative Optimization of Traffic Signals and Vehicle Speed Using a Novel Multi-agent Deep Reinforcement Learning