Control of robotic vehicles via authority-sharing between humans and robots

Marco Andreetto

Vehicle control, Authority-sharing, Robotics

An increasing number of applications requires the interaction between robotic vehicles and humans. Classical robot control applies to completely autonomous vehicles. Although this approach secures the typical precision and accuracy of robots in executing low-level tasks, it does not exploit the superior high-level cognitive abilities of humans in managing complex and unpredictable situations. We propose to improve human-robot interaction through an authority-sharing paradigm in which the control system shares the control authority with the human by combining the advantages of the low-level robot abilities and the high-level human cognition.




Activity Planning for Assistive Robotics

Paolo Bevilacqua

planning, robotics

In classic robotic applications, the motion planner generates a path to bring a robot from an initial state to a final one. When a robot is used to assist a human user in the navigation of a complex place, a more complete support for planning an activity is required (e.g. which places to visit, in which sequence) to meet user personal goals and to comply with her specific requirements. Several activity planners in the literature fulfil this goal reasoning at an abstract level. However, the generated plans are not easily translated into executable motion plans for the robot. This proposal aims to bridge this gap, through the development of a tool for activity planning of assistive robots combining a careful consideration of both the user needs and the robot dynamic and kinematic constraints.




Co-Design of Mechanics and Control Software for the Next Generation of Service Robots

Stefano Divan

Robotics, Embedded Systems, Path Following, Authority Sharing

We propose a new generation of robotic walking assistants whose mechanical and electronic components are conceived to optimize the collaboration between the robot and its users. The main goal of the system is to guide the user across complex public spaces securing a high level of safety, respect of unwritten social rules, and a smooth and non--intrusive interaction between human and machine. The perfect synergy between the robot and its user can be fulfilled by adopting holistic and multi--disciplinary approaches. The electronics of the sensors is co--designed with the data processing algorithms and the mechanics of the actuators is co--designed with the control laws. We will apply these general ideas in at least two different contexts: force sensing to estimate the torques applied by the user and change the user perceived inertia, and guidance solutions based on differential drive and/or steering wheels control. The prototype will be extensively tested with a large base of users through a rigorous application of user--centered design.





Cyber-Physical System (CPS) Architecture Exploration: An Optimization-Based Approach

Dmitrii Kirov

Cyber-Physical Systems, Embedded Systems, System Design, Optimization, Architecture Exploration

We are working on a framework for cyber-physical system architecture exploration. We formulate the exploration problem as a mapping problem, where "virtual" components are mapped into "real" components from pre-defined libraries to minimize an objective function while guaranteeing that system requirements are satisfied. The framework leverages an extensible set of patterns to enable formal, yet flexible, requirement specification, a graph-based internal representation of the system architecture, and a set of algorithms based on mixed integer linear programming to solve the mapping problem. Its effectiveness is currently demonstrated on two industrial case studies: an aircraft power distribution network and a reconfigurable automated production line. We aim to extend the framework to address a broader category of CPS, including topology synthesis and node placement in wireless networks.




Robot Localization

Valerio Magnago

My reserarch interests are focused on authonomous veichle. I am investigating state estimation of a mobile robot through sensor fusion of an arbitrary number of sensors. For example I am working on an alghorithm that aim to optimize the landmark placement.




Indoor Localization of Wheeled Robots using Multi-sensor Data Fusion with Event-based Measurements

Payam Nazemzadeh

Sensor fusion, Indoor localization, Ambient assisted living

In the era in which the robots have started to live and work everywhere and in close contact with humans, they should accurately know their own location at any time to be able to move and perform safely. In particular, large and crowded indoor environments are challenging scenarios for robots' accurate and robust localization. My research during the PhD career addressed this problem by investigating three complementary issues, i.e. improving robots self-localization through data fusion, adopting collaborative localization (e.g. using the position information from other robots) and finally optimizing the placement of landmarks in the environment once the detection range of the chosen sensors is known.




Practical Correct Multilingual Programming

Tadeus Prastowo

software engineering, model-driven engineering, formal methods, generative programming, multilingual programming

Multilingual programs are not uncommon in practice, for example, a web application being commonly written in at least three languages: SQL, Ruby, and JavaScript. Using multiple domain-specific languages (DSLs) indeed increase the quality and speed of the development and maintenance of software systems because DSLs not only encode in their syntax specific knowledge of their target problems to facilitate concise and accurate problem descriptions but also provide precise semantics of their syntax to enable static decidable formal checks, which are founded on mathematical logic. We therefore seek to make multilingual programming not only more practical but also easier to get right.




Adaptive Models for Evolving data in Internet of Things Applications

Swaytha Sasidharan

Internet of Things, Machine Learning

Smart pervasive objects are connected through Internet of Things(IoT) and have associated contextual properties and are closely linked to evolving situations in real world environments. IoT creates a distinction of a real world with objects and a corresponding virtual representation, which facilitates applications to proactively perceive, comprehend and adapt to the real world situations. This real world represents a non-stationary environment characterized by evolving changes. This thesis aims to tackle the problem of building a generic adaptive architecture that continuously reconfigures the application to reflect and respond to such changes. The component view of the architecture consists of - Learning module: models object patterns, change detection module: recognizes changes, and finally, Adaptation module: adapts to the recognized changes. The architecture is Java based with elements of database, machine learning, rule based systems integrated. This adaptive architecture will be validated in a field test scenario of a smart hospital where moving medical objects are tracked and a predictive model continuously adapts to changing movement patterns of objects.The concept is to be tested in a smart home scenario with sensors and actuators.




Bringing Probabilistic Real-Time Guarantees to the Real World

Bernardo Villalba Frias

My research focuses on the development of a stochastic analysis tool for soft real-time systems scheduled by a reservation-based algorithm and presenting correlated inter-arrival and execution times. The analysis considers, not only single-task applications but also applications composed of multiple interacting tasks running on multicore platforms.