A5036081979- AI-based Problem Solving and Planning
- Logic, Reasoning, and Knowledge
- Social Robot Interaction and HRI
- Constraint Satisfaction and Optimization
- Reinforcement Learning in Robotics
- Multi-Agent Systems and Negotiation
- Semantic Web and Ontologies
- Context-Aware Activity Recognition Systems
- Ethics and Social Impacts of AI
- Evacuation and Crowd Dynamics
- Robotics and Automated Systems
- Machine Learning and Algorithms
- Balance, Gait, and Falls Prevention
- Human-Automation Interaction and Safety
- Muscle activation and electromyography studies
- Motor Control and Adaptation
- Robot Manipulation and Learning
- Stroke Rehabilitation and Recovery
- AI in Service Interactions
- Advanced Database Systems and Queries
- Artificial Intelligence in Healthcare and Education
- Intelligent Tutoring Systems and Adaptive Learning
- Gaze Tracking and Assistive Technology
- Personal Information Management and User Behavior
- Multimodal Machine Learning Applications
Bar-Ilan University
2022-2024
The University of Texas at Austin
2020-2023
Laboratoire d'Informatique de Paris-Nord
2022-2023
Ben-Gurion University of the Negev
2016-2020
A major goal in robotics is to enable intelligent mobile robots operate smoothly shared human-robot environments. One of the most fundamental capabilities service this competent navigation “social” context. As a result, there has been recent surge research on social navigation; and especially as it relates handling conflicts between agents during navigation. These developments introduce variety models algorithms, however area inherently interdisciplinary, many relevant papers are not...
A major challenge to deploying robots widely is navigation in human-populated environments, commonly referred as social robot . While the field of has advanced tremendously recent years, fair evaluation algorithms that tackle remains hard because it involves not just robotic agents moving static environments but also dynamic human and their perceptions appropriateness behavior. In contrast, clear, repeatable, accessible benchmarks have accelerated progress fields like computer vision,...
A prevalent assumption in human-robot and human-AI teaming is that artificial teammates should be compliant obedient. In this talk, I will question by presenting the Guide Robot Grand Challenge discussing components required to design build a service robot can intelligently disobey. This challenge encompasses variety of research problems, as exemplify via three challenges: reasoning about goals other agents, choosing when interrupt, interacting tightly coupled physical environment.
Autonomous service robots in a public setting will generate hundreds of incidental human-robot encounters, yet researchers have only recently addressed this important topic earnest. In study, we hypothesized that visual indicators human control, such as leash on robot, would impact humans' perceptions the context encounters. A pilot study (n = 26) and revised 22) including semi-structured interviews 21) were conducted. The interview data suggested presence another during encounter elicited...
In exploratory domains, agents’ behaviors include switching between activities, extraneous actions, and mistakes. Such settings are prevalent in real world applications such as interaction with open-ended software, collaborative office assistants, integrated development environments. Despite the prevalence of world, there is scarce work formalizing connection high-level goals low-level behavior inferring former from latter these settings. We present a formal grammar for describing users’...
This article provides new techniques for optimizing domain design goal and plan recognition using libraries. We define two problems: Goal Recognition Design Plan Libraries (GRD-PL) (PRD). Solving the GRD-PL helps to infer which agent is trying achieve, while solving PRD can help how going achieve its goal. For each problem, we a worst-case distinctiveness measure that an upper bound on number of observations are necessary unambiguously recognize agent’s or plan. studies relationship between...
A desirable goal for autonomous agents is to be able coordinate on the fly with previously unknown teammates. Known as “ad hoc teamwork”, enabling such a capability has been receiving increasing attention in research community. One of central challenges ad teamwork quickly recognizing current plans other and planning accordingly. In this paper, we focus scenario which teammates can communicate one another, but only at cost. Thus, they must carefully balance plan recognition based...
Most approaches for goal recognition rely on specifications of the possible dynamics actor in environment when pursuing a goal. These suffer from two key issues. First, encoding these requires careful design by domain expert, which is often not robust to noise at time. Second, existing need costly real-time computations reason about likelihood each potential In this paper, we develop framework that combines model-free reinforcement learning and alleviate careful, manual design, online...
Human-exoskeleton interactions have the potential to bring about changes in human behavior for physical rehabilitation or skill augmentation. Despite significant advances design and control of these robots, their application training remains limited. The key obstacles such paradigms are prediction human-exoskeleton interaction effects selection affect behavior. In this article, we present a method elucidate behavioral system identify expert behaviors correlated with task goal. Specifically,...
In ad hoc teamwork, multiple agents need to collaborate without having knowledge about their teammates or plans a priori. A common assumption in this research area is that the cannot communicate. However, just as two random people may speak same language, autonomous also happen share communication protocol. This paper considers how such shared protocol can be leveraged, introducing means reason Communication Ad Hoc Teamwork (CAT). The goal of work enabling improved teamwork by judiciously...
Plan recognition algorithms infer agents' plans from their observed actions. Due to imperfect knowledge about the agent's behavior and environment, it is often case that there are multiple hypotheses an consistent with observations, though only one of these correct. This paper addresses problem how disambiguate between hypotheses, by querying acting agent whether a candidate plan in matches its intentions. process performed sequentially used update set possible during process. The defines...
Goal Recognition Design (GRD) is the problem of designing a domain in way that will allow easy identification agents' goals. This work extends original GRD to Plan (PRD) which task using plan libraries order facilitate fast an agent's plan. While can help explain faster goal agent trying achieve, PRD understanding how going achieve its goal. We define new measure quantifies worst-case distinctiveness given planning domain, propose method reduce it and show reduction this three domains from...
This paper presents preliminary results of our work with a major financial company, where we try to use methods plan recognition in order investigate the interactions costumer company's online interface. In this paper, present first steps integrating algorithm real-world application for detecting and analyzing costumer. It uses novel approach from bare-bone UI data, which reasons about library at lowest level define relevancy actions domain, then it perform recognition. We inference on three...
Plan recognition, activity and goal recognition all involve making inferences about other actors based on observations of their interactions with the environment agents. This sy
While human-robot interaction studies are becoming more common, quantification of the effects repeated with an exoskeleton remains unexplored. We draw upon existing literature in human skill assessment and present extrinsic intrinsic performance metrics that quantify how human-exoskeleton system's behavior changes over time. Specifically, this paper, we a new metric provides insight into kinematics associated 'successful' movements resulting richer characterization behavior. A subject study...
With robots poised to enter our daily environments, we conjecture that they will not only need work for people, but also learn from them. An active area of investigation in the robotics, machine learning, and human-robot interaction communities is design teachable robotic agents can interactively human input. To refer these research efforts, use umbrella term Human-Interactive Robot Learning (HIRL). While algorithmic solutions learning people have been investigated a variety ways, HIRL, as...