1. |
- Lamanna, Leonardo, et al.
(author)
-
Learning to Act for Perceiving in Partially Unknown Environments
- 2023
-
In: Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence (IJCAI 2023). - : International Joint Conferences on Artificial Intelligence. - 9781956792034 ; , s. 5485-5493
-
Conference paper (peer-reviewed)abstract
- Autonomous agents embedded in a physical environment need the ability to correctly perceive the state of the environment from sensory data. In partially observable environments, certain properties can be perceived only in specific situations and from certain viewpoints that can be reached by the agent by planning and executing actions. For instance, to understand whether a cup is full of coffee, an agent, equipped with a camera, needs to turn on the light and look at the cup from the top. When the proper situations to perceive the desired properties are unknown, an agent needs to learn them and plan to get in such situations. In this paper, we devise a general method to solve this problem by evaluating the confidence of a neural network online and by using symbolic planning. We experimentally evaluate the proposed approach on several synthetic datasets, and show the feasibility of our approach in a real-world scenario that involves noisy perceptions and noisy actions on a real robot.
|
|
2. |
- Lamanna, Leonardo, et al.
(author)
-
Planning for Learning Object Properties
- 2023
-
In: Proceedings of the AAAI Conference on Artificial Intelligence. - : AAAI Press. - 9781577358800 ; , s. 12005-12013
-
Conference paper (peer-reviewed)abstract
- Autonomous agents embedded in a physical environment need the ability to recognize objects and their properties from sensory data. Such a perceptual ability is often implemented by supervised machine learning models, which are pre-trained using a set of labelled data. In real-world, open-ended deployments, however, it is unrealistic to assume to have a pre-trained model for all possible environments. Therefore, agents need to dynamically learn/adapt/extend their perceptual abilities online, in an autonomous way, by exploring and interacting with the environment where they operate. This paper describes a way to do so, by exploiting symbolic planning. Specifically, we formalize the problem of automatically training a neural network to recognize object properties as a symbolic planning problem (using PDDL). We use planning techniques to produce a strategy for automating the training dataset creation and the learning process. Finally, we provide an experimental evaluation in both a simulated and a real environment, which shows that the proposed approach is able to successfully learn how to recognize new object properties.
|
|
3. |
- Bontempi, Gianluca, et al.
(author)
-
The CLAIRE COVID-19 initiative : approach, experiences and recommendations
- 2021
-
In: Ethics and Information Technology. - : Springer. - 1388-1957 .- 1572-8439. ; 23:Suppl. 1, s. 127-133
-
Journal article (peer-reviewed)abstract
- A volunteer effort by Artificial Intelligence (AI) researchers has shown it can deliver significant research outcomes rapidly to help tackle COVID-19. Within two months, CLAIRE's self-organising volunteers delivered the World's first comprehensive curated repository of COVID-19-related datasets useful for drug-repurposing, drafted review papers on the role CT/X-ray scan analysis and robotics could play, and progressed research in other areas. Given the pace required and nature of voluntary efforts, the teams faced a number of challenges. These offer insights in how better to prepare for future volunteer scientific efforts and large scale, data-dependent AI collaborations in general. We offer seven recommendations on how to best leverage such efforts and collaborations in the context of managing future crises.
|
|
4. |
- Buyukgoz, Sera, et al.
(author)
-
Two ways to make your robot proactive : Reasoning about human intentions or reasoning about possible futures
- 2022
-
In: Frontiers in Robotics and AI. - : Frontiers Media S.A.. - 2296-9144. ; 9
-
Journal article (peer-reviewed)abstract
- Robots sharing their space with humans need to be proactive to be helpful. Proactive robots can act on their own initiatives in an anticipatory way to benefit humans. In this work, we investigate two ways to make robots proactive. One way is to recognize human intentions and to act to fulfill them, like opening the door that you are about to cross. The other way is to reason about possible future threats or opportunities and to act to prevent or to foster them, like recommending you to take an umbrella since rain has been forecast. In this article, we present approaches to realize these two types of proactive behavior. We then present an integrated system that can generate proactive robot behavior by reasoning on both factors: intentions and predictions. We illustrate our system on a sample use case including a domestic robot and a human. We first run this use case with the two separate proactive systems, intention-based and prediction-based, and then run it with our integrated system. The results show that the integrated system is able to consider a broader variety of aspects that are required for proactivity.
|
|
5. |
- Faridghasemnia, Mohamadreza, 1991-, et al.
(author)
-
Towards Abstract Relational Learning in Human Robot Interaction
- 2020
-
In: CognitIve RobotiCs for intEraction (CIRCE).
-
Conference paper (peer-reviewed)abstract
- Humans have a rich representation of the entities in their environment. Entities are described by their attributes, and entities that share attributes are often semantically related. For example, if two books have "Natural Language Processing" as the value of their `title' attribute, we can expect that their `topic' attribute will also be equal, namely, "NLP". Humans tend to generalize such observations, and infer sufficient conditions under which the `topic' attribute of any entity is "NLP". If robots need to interact successfully with humans, they need to represent entities, attributes, and generalizations in a similar way. This ends in a contextualized cognitive agent that can adapt its understanding, where context provides sufficient conditions for a correct understanding. In this work, we address the problem of how to obtain these representations through human-robot interaction. We integrate visual perception and natural language input to incrementally build a semantic model of the world, and then use inductive reasoning to infer logical rules that capture generic semantic relations, true in this model. These relations can be used to enrich the human-robot interaction, to populate a knowledge base with inferred facts, or to remove uncertainty in the robot's sensory inputs.
|
|
6. |
- Gugliermo, Simona, 1995-, et al.
(author)
-
Extracting Planning Domains from Execution Traces : a Progress Report
- 2023
-
Conference paper (peer-reviewed)abstract
- One of the difficulties of using AI planners in industrial applications pertains to the complexity of writing planning domain models. These models are typically constructed by domain planning experts and can become increasingly difficult to codify for large applications. In this paper, we describe our ongoing research on a novel approach to automatically learn planning domains from previously executed traces using Behavior Trees as an intermediate human-readable structure. By involving human planning experts in the learning phase, our approach can benefit from their validation. This paper outlines the initial steps we have taken in this research, and presents the challenges we face in the future.
|
|
7. |
- Köckemann, Uwe, 1983-, et al.
(author)
-
Planning for Automated Testing of Implicit Constraints in Behavior Trees
- 2023
-
In: Proceedings of the Thirty-Third International Conference on Automated Planning and Scheduling. - : AAAI Press. ; , s. 649-658
-
Conference paper (peer-reviewed)abstract
- Behavior Trees (BTs) are a formalism increasingly used to control the execution of robotic systems. The strength of BTs resides in their compact, hierarchical and transparent representation. However, when used in practical applications transparency is often hindered by the introduction of implicit run-time relations between nodes, e.g., because of data dependencies or hardware-related ordering constraints. Manually verifying the correctness of a BT with respect to these hidden relations is a tedious and error-prone task. This paper presents a modular planning-based approach for automatically testing BTs offline at design time, to identify possible executions that may violate given data and ordering constraints and to exhibit traces of these executions to help debugging. Our approach supports both basic and advanced BT node types, e.g., supporting parallel behaviors, and can be extended with other node types as needed. We evaluate our approach on BTs used in a commercially deployed robotics system and on a large set of randomly generated trees showing that our approach scales to realistic sizes of more than 3000 nodes.
|
|
8. |
- Saffiotti, Alessandro, Professor, 1960-, et al.
(author)
-
On human-AI collaboration in artistic performance
- 2020
-
In: NeHuAI 2020: First International Workshop on New Foundations for Human-Centered AI. - : CEUR-WS. ; , s. 38-43
-
Conference paper (peer-reviewed)abstract
- Live artistic performance, like music, dance or acting, provides an excellent domain to observe and analyze the mechanisms of human-human collaboration. In this note, we use this domain to study human-AI collaboration. We propose a model for collaborative artistic performance, in which an AI system mediates the interaction between a human and an artificial performer. We then instantiate this model in three case studies involving different combinations of human musicians, human dancers, robot dancers, and a virtual drummer. All case studies have been demonstrated in public live performances involving improvised artistic creation, with audiences of up to 250 people. We speculate that our model can be used to enable human-AI collaboration beyond the domain of artistic performance.
|
|
9. |
- Thörn, Oscar, 1994-, et al.
(author)
-
Human-Robot Artistic Co-Creation : a Study in Improvised Robot Dance
- 2020
-
In: 2020 29th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN). - : IEEE. - 9781728160757 ; , s. 845-850
-
Conference paper (peer-reviewed)abstract
- Joint artistic performance, like music, dance or acting, provides an excellent domain to observe the mechanisms of human-human collaboration. In this paper, we use this domain to study human-robot collaboration and co-creation. We propose a general model in which an AI system mediates the interaction between a human performer and a robotic performer. We then instantiate this model in a case study, implemented using fuzzy logic techniques, in which a human pianist performs jazz improvisations, and a robot dancer performs classical dancing patterns in harmony with the artistic moods expressed by the human. The resulting system has been evaluated in an extensive user study, and successfully demonstrated in public live performances.
|
|
10. |
- Tomic, Stevan, 1981-
(author)
-
Human Norms for Robotic Minds
- 2022
-
Doctoral thesis (other academic/artistic)abstract
- Interactions within human societies are usually regulated by social norms. When humans fail to cope with norms, their behavior may be perceived as antisocial, amusing, provocative, dumb, or even uncanny, depending on the intention ascribed to the actor by the observer. Robots cannot yet follow social norms, and often their behavior is judged in such a manner. We claim that acceptance of robots into human societies will critically depend on their ability to represent, reason about, and learn social norms. How to provide these abilities is the main problem addressed in the thesis.We formalize the notion of social context as an institution that encapsulates a set of abstract social norms. Then, we connect these abstract norms to physical execution through an explicit notion of grounding, that puts together three levels of representation: (1) a formalization of normative structures in human-readable terms, (2) a mapping of these formal structures to models of execution in the physical world, and (3) a vector space representation of all these elements suitable for machine learning algorithms. Given this background, we identify computational problems central to reasoning and learning with norms, and provide solutions to several of these problems.We start by considering two reasoning problems: verification, i.e., how to verify whether or not a physical execution (or its interpretation) adheres to a set of norms; and planning, i.e., how to generate plans adherent to norms. We address these problems by reducing them to known problems with known solutions. Specifically, the verification problem is reduced to a constraint satisfaction problem (CSP), which, in turn, allows us to address planning as a meta-CSP.We then address two problems related to learning with norms: how to learn policies that generate adherent trajectories; and how to re-use learned policies across physical domains. To do so, we translate the elements of our framework into a vector-space format. We address the former problem by using norm verification mechanisms to guide a reinforcement learning agent. To address the latter problem we break it into two sub-problems: interpreting the current situation in a context, and performing the actual (reinforcement) learning. That puts us in a unique position of combining reasoning and learning by searching in the space of groundings and re-usable policies.We evaluate all our solutions with use-case experiments, both on real robots and on simulated agents. Exploring the concepts of abstraction and of interpretation in physical execution touches upon some general questions about intelligence and computation, hence we complement the technical contributions with a discussion of our work from the perspective of Cognitive Science.
|
|