arXiv subjects

Sara Honarvar

Publications and source records attributed to Sara Honarvar.

3 recordsLinked to original sources

arXiv2024

Learning Optimal Interaction Weights in Multi-Agents Systems

This paper presents a spatio-temporal inverse optimal control framework for understanding interactions in multi-agent systems (MAS). We employ a graph representation approach and model the dynamics of interactions between agents as state-dependent edge weights in a consensus algorithm, incorporating both spatial and temporal dynamics. Our method learns these edge weights from trajectory observations, such as provided by expert demonstrations, which allows us to capture the complexity of nonlinear, distributed interaction behaviors. We derive necessary and sufficient conditions for the optimality of these interaction weights, explaining how the network topology affects MAS coordination. The proposed method is demonstrated on a multi-agent formation control problem, where we show its effectiveness in recovering the interaction weights and coordination patterns from sample trajectory data.

eess.SY↗

arXiv2024

Geometric Graph Neural Network Modeling of Human Interactions in Crowded Environments

Modeling human trajectories in crowded environments is challenging due to the complex nature of pedestrian behavior and interactions. This paper proposes a geometric graph neural network (GNN) architecture that integrates domain knowledge from psychological studies to model pedestrian interactions and predict future trajectories. Unlike prior studies using complete graphs, we define interaction neighborhoods using pedestrians' field of view, motion direction, and distance-based kernel functions to construct graph representations of crowds. Evaluations across multiple datasets demonstrate improved prediction accuracy through reduced average and final displacement error metrics. Our findings underscore the importance of integrating domain knowledge with data-driven approaches for effective modeling of human interactions in crowds.

cs.LG↗

arXiv2021

Modeling Human-Human Collaboration: A Connection Between Inter-Personal Motor Synergy and Consensus Algorithms

Many day-to-day activities involve people working collaboratively toward reaching a desired outcome. Previous research in motor control and neuroscience have proposed inter-personal motor synergy (IPMS) as a mechanism of collaboration between people, referring to the idea of how two or more people may work together "as if they were one" to coordinate their motion. In motor control literature, uncontrolled manifold (UCM) is used for quantifying IPMS. According to this approach, coordinated motion is achieved through stabilization of a performance variable (e.g., an output in a collaborative output tracking task). We show that the UCM approach is closely related to the well-studied consensus approach in multi-agent systems that concerns processes by which a set of interacting agents agree on a shared objective. To explore the connection between these two approaches, in this paper, we provide a control-theoretic model that represents the systems-level behaviors in a collaborative task. In particular, we utilize the consensus protocol and show how the model can be systematically tuned to reproduce the behavior exhibited by human-human collaboration experiments. We discuss the association between the proposed control law and the UCM approach and validate our model using experimental results previously collected from an inter-personal finger force production task.

eess.SY↗