Schedule

9:15

Workshop presentation and initial discussion

9:30

Paolo Robuffo Giordano

Formation Control and Localization with Onboard Sensing and Local Communication

10:00

Vadim Indelman

Advances in Computationally Efficient and Robust (Multi-Robot) Belief Space Planning in High-Dimensional State Spaces

10:30

Dimos Dimarogonas

Distributed hybrid control of multi-robot systems under spatio-temporal specifications

11:00-11:30

Coffee Break

11:30

Alberto Sanfeliu

Multi-Robot Person Searching and Following

12:00

Marco Karrer, Patrik Schmuck and Margarita Chli

Towards advanced visual perception for robotic teams

12:30

Poster Spotlights 1

13:00

Poster discussion 1

13:30-14:30

Lunch Break

14:30

Andrea Gasparri

Stable Coordinated Motion for Teams with Limited Fields of View 

15:00

Alcherio Martinoli

Fluid-Mediated Stochastic Self-Assembly: Towards Bridging Centimetric and Submillimetric Scales

15:30

Mac Schwager

Ants Don't Use WiFi: Enabling Robotic Agents to Collaborate and Compete through Perception 

16:00

Poster Spotlights 2

16:30-17:00

Coffee Break with poster discussion

17:00

Poster discussion 2

17:30

Workshop closure


Talk abstracts


Paolo Robuffo Giordano

Formation Control and Localization with Onboard Sensing and Local Communication

Abstract: Research on multi-robot systems has flourished over the last decades with a number of theoretical and experimental results also made possible by the constant technological advancements in onboard sensing, communication and computing power. A scenario that still motivates considerable research efforts is that of decentralized formation control of multiple mobile robots based on only local (onboard) sensing and communication, with the aim of deploying highly autonomous robot teams in "non-trivial" environments (e.g., inside buildings, underwater, underground, or even in deep space) where centralized measuring/communication facilities (such as GPS) are not available.

In this talk, I will review several recent results involving decentralized formation control and localization for groups of robots only relying on onboard sensing and (when necessary) local communication. The focus will be on distance sensors and bearing sensors (representative of onboard cameras). I will start considering basic formation control problems (reaching a desired formation) and then move on to more complex mission requirements, involving active maintenance of global properties, and active estimation for improved localization performance. Experiments with quadrotor UAVs will finally help assessing how the various theoretical and algorithmic contributions actually perform in realistic settings.


Vadim Indelman

Advances in Computationally Efficient and Robust (Multi-Robot) Belief Space Planning in High-Dimensional State Spaces

Abstract: In this talk I will provide an overview of recent approaches developed in my group to address single and multi-robot belief space planning (BSP) in high-dimensional state spaces while considering, as an application, autonomous operation in unknown/uncertain environments (e.g. active SLAM). Specifically, these approaches focus on (i) computationally efficient calculation of an information-theoretic cost (e.g. entropy) while re-using calculations and avoiding explicit belief propagation for different candidate (non-myopic) actions; (ii) simplifying the BSP problem via sparsification and resorting to a topological space; and (iii) coping with ambiguous data association via multi-modal BSP, thereby enabling active disambiguation capabilities.


Dimos Dimarogonas

Distributed hybrid control of multi-robot systems under spatio-temporal specifications

Multi-robot task planning and control under temporal logic specifications has been gaining increasing attention in recent years due to its applicability among others in autonomous systems, manufacturing systems, service robotics and assisted living. Initial approaches considered qualitative logics, such as Linear Temporal Logic, whose automata representation facilitates the direct use of model checking tools for correct-by-design control synthesis. In many real world applications however, there is a need to quantify spatial and temporal constraints, e.g., in order to include deadlines and separation assurance bounds. This led to the use of quantitative logics, such as Metric Interval and Signal Temporal Logic, to impose such spatiotemporal constraints. However, the lack of automata representations for such specifications hinders the direct use of model checking tools. Motivated by this, the use of transient control methodologies that fulfill the aforementioned qualitative constraints becomes evident. In this talk, we review some of our recent results in applying transient control techniques, and in particular Model Predictive Control, Barrier Certificates based design and Prescribed Performance Control, to distributed multi-robot task planning under spatiotemporal specifications. We consider the case of infeasible specifications and propose a least violating control strategy as a remedy. The results are supported by relevant experimental validations.


Alberto Sanfeliu

Multi-Robot Person Searching and Following

Abstract: We will present diverse techniques developed in the Institut de Robotica I Informatica Industrial (CSIC-UPC) using multi-robots, to search and follow people in urban spaces where the are static obstacles (columns, rooms, trees, etc.) and dynamic obstacles, for example people moving around. The robots have to use multi-exploration techniques to find the hidden person, and once it is found, to follow it. The robots have to handle with partial occlusions due to other people moving, false person detections, or sensor failures.


Marco Karrer, Patrik Schmuck and Margarita Chli

Towards advanced visual perception for robotic teams

Abstract: Robotic collaboration promises increased robustness and efficiency with multi-agent SLAM right at the core of enabling collaboration, such that each agent can co-localize in and build a map of the workspace. With some of the biggest challenges lying with robust communication, efficient data management and effective sharing of information amongst the agents, in this talk, we will discuss recent work at V4RL (www.v4rl.ethz.ch) in this direction touching on vision-based collaboration both peer-to-peer as well as centralized architectures for multi-robot SLAM.


Andrea Gasparri

Stable Coordinated Motion for Teams with Limited Fields of View

Abstract: In this work, we address the problem of coordinating the motion of a team of robots with limited fields of view (FOVs). In this regard our contribution is threefold: i) first, we propose a potential-based stable coordinated motion framework for multi-robot systems with heterogeneous disk-based interactions; ii) then, we introduce a switching control mechanism for controlling collaborative behaviour, collision avoidance, and topology recovery with the support of a centralized planner; iii) finally, we generalize the proposed coordination framework to consider directed interactions encoded by sensors with limited fields of view. Experimental validations are provided to corroborate the theoretical findings.


Alcherio Martinoli

Fluid-Mediated Stochastic Self-Assembly: Towards Bridging Centimetric and Submillimetric Scales

Abstract: Miniature robots at centimeter scale can be effective demonstrators: they can be designed and manufactured leveraging off-the-shelf components and standard mechatronic recipes. Unfortunately, the application areas for this scale are limited while many potential applications are available at the submillimeter scale. Devices at the submillimeter scale cannot be endowed with similar resources as their centimetric counterpart as the manufacturing technology at this scale is still very expensive as well as customized, and multiple functionalities are difficult to integrate in a single device with canonical top-down manufacturing techniques. One of the promising techniques to manufacture more complex microrobots is to leverage existing MEMS achieving different functionalities, produced with standard micromachining procedures, and use them as building blocks for a fluid-mediated self-assembling process. In order to efficiently guide the self-assembly process, stochastic modeling and control techniques developed for centimeter devices can help. In this talk, I will illustrate two of these techniques, the first one being essentially centralized while the second one distributed; they are both perception-driven and involve some simple planning in the guidance of the self-assembling process towards the target shape. The first method leverages a vision-based closed-loop framework able to both automatically create models at multiple abstraction levels and optimize the agitation of the liquid in which the self-assembly process takes place. The second method relies instead on programmable rulesets which can be downloaded on robotic modules able to communicate with neighbors and control their on-board latching properties. A combination of such methods could be potentially deployable at submillimeter level in the future.


Mac Schwager

Ants Don't Use WiFi: Enabling Robotic Agents to Collaborate and

Compete through Perception

Abstract: In the animal world there is no WiFi---agents collaborate and compete by sensing and predicting the actions of teammates, rivals, predators, and prey.  Likewise, in the engineered world, many of the most promising applications for autonomous robots require them to interact with other agents in the world by sensing and predicting their actions.  Autonomous driving in traffic, collision avoidance for UAVs, and human-robot teaming are key examples where a wireless network either cannot exist, or will not exist for some time.  In competitive scenarios, such as racing or pursuit-evasion, agents would not want to communicate even if they could.  In this talk I will describe two recent examples from my lab of algorithms enabling multiple robotic

agents to interact, both collaboratively and competitively, without a communication network.  First, I will discuss a communication-free multi-robot manipulation algorithm by which robots cooperate to transport a payload by sensing the motion of the payload.  Second, I will present a game theoretic receding horizon control algorithm for

autonomous drone racing, in which drones sense each other's position with a monocular camera.  I will show results from hardware experiments with ground robots, and quadrotor UAVs collaborating and competing in the scenarios above.