Elisa Tosello
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Elisa Tosello has a total of 57 co-authors. Here are the top co-authors by number of shared publications:
- Emanuele Menegatti
University of Padua
9 shared publications Enrico Pagello
Intelligent Autonomous Systems Lab (IAS-Lab), Department of Information Engineering, University of Padova, Padova, Italy
4 shared publications- Nicola Castaman
Department of Information Engineering, University of Padova, Via Gradenigo 6/B, 35131, Padova, Italy and IT+Robotics srl, Contr‘ Valmerlara 21, 36100, Vicenza, Italy
4 shared publications - Andrea Micheli
Fondazione Bruno Kessler
3 shared publications Roberto Bortoletto
Intelligent Autonomous Systems Lab (IAS-Lab), Department of Information Engineering, University of Padova, Padova, Italy
3 shared publications- Stefano Ghidoni
University of Padova
3 shared publications
Latest Publications
- Open Access
- 28 September 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Autonomous and Adaptive Systems
Incremental Adaptation of Human-Robot Collaborative Missions using Hybrid Task Planning
As robots increasingly collaborate with humans on complex missions, coordinating tasks under functional and non-functional constraints becomes more challenging, particularly in dynamically changing environments. Uncertainty arising from unpredictable human behaviour, environmental variability, and system failures further complicates the synthesis of robust task plans that ensure mission success while optimising cost, safety, and performance. Further, the presence of multiple, often conflicting objectives, such as maximising the probability of mission success while minimising expected execution cost, together with probabilistic action outcomes, exacerbates the combinatorial complexity of providing formal guarantees. Addressing these challenges entails that effective planning approaches must combine scalability with rigorous verification and provide adaptive capabilities to cope with runtime disruptions. We introduce ARCH (Adaptive Robot–Human Collaboration using Hybrid Planning), an approach for synthesising verified and adaptable task plans for complex Cyber-Physical-Human Systems (CPHS). ARCH integrates off-the-shelf numerical planning, probabilistic model checking, and multi-objective evolutionary algorithms to generate Pareto-optimal sets of task plans that satisfy probabilistic requirements. To support efficient adaptation, ARCH reuses previously synthesised plans that remain feasible after changes, seeding the evolutionary search and guiding it towards promising regions of the solution space. We evaluate ARCH using two industrial CPHS case studies and conduct an empirical analysis of its effectiveness and scalability. The results demonstrate that seeding can significantly improve search efficiency during adaptation and that ARCH scales to problem instances where traditional probabilistic planning approaches become impractical due to state explosion.
- 28 September 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Autonomous and Adaptive Systems
Open AccessIncremental Adaptation of Human-Robot Collaborative Missions using Hybrid Task Planning
As robots increasingly collaborate with humans on complex missions, coordinating tasks under functional and non-functional constraints becomes more challenging, particularly in dynamically changing environments. Uncertainty arising from unpredictable human behaviour, environmental variability, and system failures further complicates the synthesis of robust task plans that ensure mission success while optimising cost, safety, and performance. Further, the presence of multiple, often conflicting objectives, such as maximising the probability of mission success while minimising expected execution cost, together with probabilistic action outcomes, exacerbates the combinatorial complexity of providing formal guarantees. Addressing these challenges entails that effective planning approaches must combine scalability with rigorous verification and provide adaptive capabilities to cope with runtime disruptions. We introduce ARCH (Adaptive Robot–Human Collaboration using Hybrid Planning), an approach for synthesising verified and adaptable task plans for complex Cyber-Physical-Human Systems (CPHS). ARCH integrates off-the-shelf numerical planning, probabilistic model checking, and multi-objective evolutionary algorithms to generate Pareto-optimal sets of task plans that satisfy probabilistic requirements. To support efficient adaptation, ARCH reuses previously synthesised plans that remain feasible after changes, seeding the evolutionary search and guiding it towards promising regions of the solution space. We evaluate ARCH using two industrial CPHS case studies and conduct an empirical analysis of its effectiveness and scalability. The results demonstrate that seeding can significantly improve search efficiency during adaptation and that ARCH scales to problem instances where traditional probabilistic planning approaches become impractical due to state explosion.
- Open Access
- 13 July 2026
- Published by Elsevier in SSRN Electronic Journal
Interactive Human-AI Scheduling in Smart Manufacturing: a Sliding Work Sharing Framework with Digital Twin Simulation
With the growing need for flexible and adaptive production systems, this paper presents an Artificial Intelligence (AI)-driven framework for manufacturing scheduling that supports human operators in advanced decision-making. The framework enables a context-sensitive allocation of decision authority between human and AI actors through a sliding work sharing paradigm while ensuring transparent human oversight.The framework integrates an Optimized Decision-Maker, which automatically generates feasible schedules based on activities, resources, and constraints, with a Digital Twin that simulates schedule execution and computes relevant Key Performance Indicators (KPIs). These KPIs are analyzed by a Sliding Work Sharing (SWS) component that leverages fuzzy-rule-based reasoning to determine the appropriate level of human involvement in validating or refining the generated decision.Through a Graphical User Interface, human users can monitor KPIs and SWS recommendations, perform what-if analyses, and introduce additional constraints. This interactive process operates in real time, enabling schedules to adapt to changing operational conditions.Developed within the EU-funded Ai4Work project, the framework supports line managers in scheduling replenishment activities for the E-CO Flex box-forming machine by I.M.A. S.p.A.. In the presented case study, plan robustness, reflecting the line's ability to sustain throughput, and worker well-being are considered the primary KPIs, balancing production efficiency with human-centered objectives.Obtained results demonstrate effective integration of human input and positive user acceptance, highlighting the framework's potential to foster productive human–AI collaboration in smart manufacturing.
- 13 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessInteractive Human-AI Scheduling in Smart Manufacturing: a Sliding Work Sharing Framework with Digital Twin Simulation
With the growing need for flexible and adaptive production systems, this paper presents an Artificial Intelligence (AI)-driven framework for manufacturing scheduling that supports human operators in advanced decision-making. The framework enables a context-sensitive allocation of decision authority between human and AI actors through a sliding work sharing paradigm while ensuring transparent human oversight.The framework integrates an Optimized Decision-Maker, which automatically generates feasible schedules based on activities, resources, and constraints, with a Digital Twin that simulates schedule execution and computes relevant Key Performance Indicators (KPIs). These KPIs are analyzed by a Sliding Work Sharing (SWS) component that leverages fuzzy-rule-based reasoning to determine the appropriate level of human involvement in validating or refining the generated decision.Through a Graphical User Interface, human users can monitor KPIs and SWS recommendations, perform what-if analyses, and introduce additional constraints. This interactive process operates in real time, enabling schedules to adapt to changing operational conditions.Developed within the EU-funded Ai4Work project, the framework supports line managers in scheduling replenishment activities for the E-CO Flex box-forming machine by I.M.A. S.p.A.. In the presented case study, plan robustness, reflecting the line's ability to sustain throughput, and worker well-being are considered the primary KPIs, balancing production efficiency with human-centered objectives.Obtained results demonstrate effective integration of human input and positive user acceptance, highlighting the framework's potential to foster productive human–AI collaboration in smart manufacturing.
- 6 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics & Automation Magazine
- Vol. 31 (1) , 72-83
Opportunistic (Re)planning for Long-Term Deep-Ocean Inspection: An Autonomous Underwater Architecture
Robots are increasingly used in subsea environments because of their positive impact on human safety and operational capabilities in the deep ocean. However, achieving full autonomy remains challenging because of the extreme conditions they encounter. In this context, we propose an autonomous underwater architecture for long-term deep-ocean inspection that robustly plans activities and efficiently deliberates with no human help. It combines the innovative Saipem’s Hydrone-R subsea vehicle with an advanced planning architecture, resulting in a robot that autonomously perceives its surroundings, plans a mission, and adapts in real time to contingencies and opportunities. After describing the robot hardware, we present the technological advancements achieved in building its software, along with several compelling use cases. We explore scenarios where the robot conducts long-term underwater missions operating under resource constraints while remaining responsive to opportunities, such as new inspection points. Our solution gained significant reliability and acceptance within the oil and gas community as evidenced by its current deployment on a real field in Norway.
- 6 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics & Automation Magazine
- Vol. 31 (1) , 72-83
Opportunistic (Re)planning for Long-Term Deep-Ocean Inspection: An Autonomous Underwater Architecture
Robots are increasingly used in subsea environments because of their positive impact on human safety and operational capabilities in the deep ocean. However, achieving full autonomy remains challenging because of the extreme conditions they encounter. In this context, we propose an autonomous underwater architecture for long-term deep-ocean inspection that robustly plans activities and efficiently deliberates with no human help. It combines the innovative Saipem’s Hydrone-R subsea vehicle with an advanced planning architecture, resulting in a robot that autonomously perceives its surroundings, plans a mission, and adapts in real time to contingencies and opportunities. After describing the robot hardware, we present the technological advancements achieved in building its software, along with several compelling use cases. We explore scenarios where the robot conducts long-term underwater missions operating under resource constraints while remaining responsive to opportunities, such as new inspection points. Our solution gained significant reliability and acceptance within the oil and gas community as evidenced by its current deployment on a real field in Norway.
- 8 April 2022
- Published by Springer Nature
- p. 532-550
Robotic Arm Control and Task Training Through Deep Reinforcement Learning
- 8 April 2022
- Published by Springer Nature
- p. 532-550
Robotic Arm Control and Task Training Through Deep Reinforcement Learning
- 8 April 2022
- Published by Springer Nature
- p. 563-579
A Planning Domain Definition Language Generator, Interpreter, and Knowledge Base for Efficient Automated Planning
- 8 April 2022
- Published by Springer Nature
- p. 563-579
A Planning Domain Definition Language Generator, Interpreter, and Knowledge Base for Efficient Automated Planning
- 15 March 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Human-Machine Systems
- Vol. 52 (3) , 410-422
Shared Control in Robot Teleoperation With Improved Potential Fields
In shared control teleoperation, the robot assists the user in accomplishing the desired task. Rather than simply executing the user’s command, the robot attempts to integrate it with information from the environment, such as obstacle and/or goal locations, and it modifies its behavior accordingly. In this article, we propose a real-time shared control teleoperation framework based on an artificial potential field approach improved by the dynamic generation of escape points around the obstacles. These escape points are virtual attractive points in the potential field that the robot can follow to overcome the obstacles more easily. The selection of which escape point to follow is done in real time by solving a soft-constrained problem optimizing the reaching of the most probable goal, estimated from the user’s action. Our proposal has been extensively compared with two state-of-the-art approaches in a static cluttered environment and a dynamic setup with randomly moving objects. Experimental results showed the efficacy of our method in terms of quantitative and qualitative metrics. For example, it significantly decreases the time to complete the tasks and the user’s intervention, and it helps reduce the failure rate. Moreover, we received positive feedback from the users that tested our proposal. Finally, the proposed framework is compatible with both mobile and manipulator robots.
- 15 March 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Human-Machine Systems
- Vol. 52 (3) , 410-422
Shared Control in Robot Teleoperation With Improved Potential Fields
In shared control teleoperation, the robot assists the user in accomplishing the desired task. Rather than simply executing the user’s command, the robot attempts to integrate it with information from the environment, such as obstacle and/or goal locations, and it modifies its behavior accordingly. In this article, we propose a real-time shared control teleoperation framework based on an artificial potential field approach improved by the dynamic generation of escape points around the obstacles. These escape points are virtual attractive points in the potential field that the robot can follow to overcome the obstacles more easily. The selection of which escape point to follow is done in real time by solving a soft-constrained problem optimizing the reaching of the most probable goal, estimated from the user’s action. Our proposal has been extensively compared with two state-of-the-art approaches in a static cluttered environment and a dynamic setup with randomly moving objects. Experimental results showed the efficacy of our method in terms of quantitative and qualitative metrics. For example, it significantly decreases the time to complete the tasks and the user’s intervention, and it helps reduce the failure rate. Moreover, we received positive feedback from the users that tested our proposal. Finally, the proposed framework is compatible with both mobile and manipulator robots.
- 3 November 2020
- Published by Taylor & Francis in Advanced Robotics
- Vol. 35 (1) , 1-18
RUR53: an unmanned ground vehicle for navigation, recognition, and manipulation
This paper proposes RUR53: an Unmanned Ground Vehicle able to navigate through, identify, and reach areas of interest. There, it can recognize, localize, and manipulate work tools to perform both indoor and outdoor complex tasks. Indeed, a wide range of sensors composes the robot and enables it to perceive vast workspaces, reach distant targets, and face the uncertainties of the real world. Precise object detection is also guaranteed, essential to manipulate objects of different shapes and materials. Moreover, a customized 3-finger gripper makes the gripping mode suitable for any lightweight object. Two modalities are proposed: autonomous and teleoperated, letting both unskilled and skilled human operators easily adapt the system to complete personalized tasks. The paper exhaustively describes RUR53 architecture and demonstrates its good performance while executing both indoor and outdoor navigation and manipulation tasks. A specific case study is described where the proposed modular architecture allows to easily switch to a semi-teleoperated mode: the 2017 Mohamed Bin Zayed International Robotics Challenge, where our team ranked third in the Grand Challenge in collaboration with the Czech Technical University in Prague, the University of Pennsylvania, and the University of Lincoln (UK). GRAPHICAL ABSTRACT
- 3 November 2020
- Published by Taylor & Francis in Advanced Robotics
- Vol. 35 (1) , 1-18
RUR53: an unmanned ground vehicle for navigation, recognition, and manipulation
This paper proposes RUR53: an Unmanned Ground Vehicle able to navigate through, identify, and reach areas of interest. There, it can recognize, localize, and manipulate work tools to perform both indoor and outdoor complex tasks. Indeed, a wide range of sensors composes the robot and enables it to perceive vast workspaces, reach distant targets, and face the uncertainties of the real world. Precise object detection is also guaranteed, essential to manipulate objects of different shapes and materials. Moreover, a customized 3-finger gripper makes the gripping mode suitable for any lightweight object. Two modalities are proposed: autonomous and teleoperated, letting both unskilled and skilled human operators easily adapt the system to complete personalized tasks. The paper exhaustively describes RUR53 architecture and demonstrates its good performance while executing both indoor and outdoor navigation and manipulation tasks. A specific case study is described where the proposed modular architecture allows to easily switch to a semi-teleoperated mode: the 2017 Mohamed Bin Zayed International Robotics Challenge, where our team ranked third in the Grand Challenge in collaboration with the Czech Technical University in Prague, the University of Pennsylvania, and the University of Lincoln (UK). GRAPHICAL ABSTRACT
- Open Access
Robotic Arm Control and Task Training through Deep Reinforcement Learning
This paper proposes a detailed and extensive comparison of the Trust Region Policy Optimization and DeepQ-Network with Normalized Advantage Functions with respect to other state of the art algorithms, namely Deep Deterministic Policy Gradient and Vanilla Policy Gradient. Comparisons demonstrate that the former have better performances then the latter when asking robotic arms to accomplish manipulation tasks such as reaching a random target pose and pick &placing an object. Both simulated and real-world experiments are provided. Simulation lets us show the procedures that we adopted to precisely estimate the algorithms hyper-parameters and to correctly design good policies. Real-world experiments let show that our polices, if correctly trained on simulation, can be transferred and executed in a real environment with almost no changes.
Open AccessRobotic Arm Control and Task Training through Deep Reinforcement Learning
This paper proposes a detailed and extensive comparison of the Trust Region Policy Optimization and DeepQ-Network with Normalized Advantage Functions with respect to other state of the art algorithms, namely Deep Deterministic Policy Gradient and Vanilla Policy Gradient. Comparisons demonstrate that the former have better performances then the latter when asking robotic arms to accomplish manipulation tasks such as reaching a random target pose and pick &placing an object. Both simulated and real-world experiments are provided. Simulation lets us show the procedures that we adopted to precisely estimate the algorithms hyper-parameters and to correctly design good policies. Real-world experiments let show that our polices, if correctly trained on simulation, can be transferred and executed in a real environment with almost no changes.
- 25 December 2019
- Published by Springer Nature in Advances in Intelligent Systems and Computing
- p. 107-119
Teaching Robot Programming for Industry 4.0
- 25 December 2019
- Published by Springer Nature in Advances in Intelligent Systems and Computing
- p. 107-119
Teaching Robot Programming for Industry 4.0
- 1 October 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 486-492
Robot Task Planning via Deep Reinforcement Learning: a Tabletop Object Sorting Application
- 1 October 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 486-492
Robot Task Planning via Deep Reinforcement Learning: a Tabletop Object Sorting Application