HUMAN-AI TEAMING
Lab

Welcome to the Human-AI Teaming (HAT) lab, in the Department of Informatics at King’s College London!

The vision of the HAT lab is that humans should work together with the AI as a team.

To achieve this, we focus on new research and challenges around Safe, Trusted and Explainable AI, that will allow humans to trust what the AI systems are about to do and to interact with the AI systems to co-create solutions. We believe that top solutions are found by leveraging human expertise and domain knowledge together with the power of cutting-edge research in AI.

We focus on AI Planning and Machine/Reinforcement Learning, and the main application area is the control and optimisation of robotics and autonomous systems, where AI is used to control (teams of) moving robots as well as to provide robots with more autonomy (for example in manufacturing, space robotics, logistics, satellites as well as robots in extreme environments).

In this context, it is estimated that the average ratio of humans supervising robots working autonomously is 5:1. It is now time to reverse the ratio, to have 1 human supervising (at least) 5 robots.

And this is the goal of the HAT lab!


We have a rich portfolio of funded projects and collaborations with many academic and non-academinc partners, and we are always looking for new partnerships for impactful research.

We are very pleased to offer the community the open source software developed by our team:

Latest news

  • New Book on PDDL
    An Introduction to Planning Domain Definition Language (PDDL) (Read more)
  • New papers on Explainable AI Planning!
    Read them now in the publications section.
  • PhD studentship available!
    Get in touch if you are interested in joining the HAT lab for a PhD!
  • New PostDoc position available!
    We are seeking a full-time PostDoc in AI and Robotics. The post is for up to 36 months. (Link for applications)
  • New Workshops in Explainable AI
    Workshop on Explainable AI at IJCAI-19 (Read more).
    Workshop on Explainable Planning at ICAPS-19 (Read more)
  • New CDT in Safe and Trusted AI!
    Apply for a PhD in the new Centre for Doctoral Training in Safe and Trusted AI! (Read more)
  • New AFOSR grant!
    New project funded by AFOSR on Explaining the Space of Plans (Read more)
  • New version of ROSPlan
    A new version of ROSPlan has been realed, together with new tutorials (Read more)
  • New InnovateUK grant!
    New project funded by InnovateUK on Intelligent Situational Awareness Platform (Read more)

The team

All the work in the team is only possible thanks to the teamwork with the brilliant Postdocs and PhD Students. Many thanks to them!

Daniele Magazzeni

Daniele Magazzeni

Reader

Gerard Canal

Gerard Canal

Postdoc

Senka Krivić

Senka Krivić

Postdoc

Dorian Buksz

Dorian Buksz

PhD Student

Anna Collins

Anna Collins

PhD Student

Sophia Kalanovska

Sophia Kalanovska

PhD Student

Benjamin Krarup

Benjamin Krarup

PhD Student

Lin Li

Lin Li

PhD Student

Parisa Zehtabi

Parisa Zehtabi

PhD Student

Associates

Michael Cashmore

Michael Cashmore

Chancellor’s Fellow

University of Strathclyde

Oscar Lima

Oscar Lima

Researcher at DFKI

German Research Center for AI

We are looking for new students to join the team. Email Dan if interested.

Projects

A list of projects we are currently involved with.

Trust in Human-Machine Partnership (ThUMP)

EPSRC/UKRI (£1.2m / 2019-2022)

Explaining the Space of Plans

AFOSR ($1m / 2018-2023)

Intelligent Situational Awareness Platform

InnovateUK (£130k / 2018-2019)

A Realtime Autonomous Robot Navigation Framework

Korea Evaluation Institute of Industrial Technology:
2018-2021

King’s/NASA Collaboration on Planning Technologies

King’s Impact Acceleration Grant

Publications

Our latest publications. A complete list can be found here.

  • Planning for Hybrid Systems via Satisfiability Modulo Theories. Michael Cashmore, Daniele Magazzeni and Parisa Zehtabi; Journal of Artificial Intelligence Research (JAIR), , (to appear), 2020.

    Abstract:
    Planning for hybrid systems is important for dealing with real-world applications, and PDDL+ supports this representation of domains with mixed discrete and continuous dynamics. In this paper we present a new approach for planning for hybrid systems, based on encoding the planning problem as a Satisfiability Modulo Theories (SMT) formula. This is the first SMT encoding that can handle the whole set of PDDL+ features (including processes and events), and is implemented in the planner SMTPlan. SMTPlan not only covers the full semantics of PDDL+, but can also deal with non-linear polynomial continuous change without discretization. This allows it to generate plans with non-linear dynamics that are correct-by-construction. The encoding is based on the notion of happenings, and can be applied on domains with nonlinear continuous change. We describe the encoding in detail and provide in-depth examples. We apply this encoding in an iterative deepening planning algorithm. Experimental results show that the approach dramatically outperforms existing work in finding plans for PDDL+ problems. We also present experiments which explore the performance of the proposed approach on temporal planning problems, showing that the scalability of the approach is limited by the size of the discrete search space. We further extend the encoding to include planning with control parameters. The extended encoding allows the definition of actions to include infinite domain parameters, called control parameters. We present experiments on a set of problems with control parameters to demonstrate the positive effect they provide to the approach of planning via SMT.
    @article{jariSMT_2020,
    author = "Cashmore, Michael and Magazzeni, Daniele and Zehtabi, Parisa",
    title = "{Planning for Hybrid Systems via Satisfiability Modulo Theories}",
    journal = "Journal of Artificial Intelligence Research (JAIR)",
    pages = "(to appear)",
    year = "2020"
    }
  • A New Approach to Plan-Space Explanation: Analyzing Plan-Property Dependencies in Oversubscription Planning. Rebecca Eiffer, Michael Cashmore, Joerg Hoffmann, Daniele Magazzeni and Marcel Steinmetz; Proceedings of AAAI Conference on Artificial Intelligence (AAAI 2020), 2020.

    Abstract:
    In many usage scenarios of AI Planning technology, users will want not just a plan pi but an explanation of the space of possible plans, justifying pi. In particular, in oversubscription planning where not all goals can be achieved, users may ask why a conjunction A of goals is not achieved by pi. We propose to answer this kind of question with the goal conjunctions B excluded by A, i.e., that could not be achieved were A enforced. We formalize this approach in terms of plan-property dependencies, where plan properties are propositional formulas over the goals achieved by a plan, and dependencies are entailment relations in plan space. We focus on entailment relations of the form A ==> not B, and devise analysis techniques globally identifying all such relations, or locally identifying the implications of a single given plan property (user question) A. We show how, via compilation, one can analyze dependencies between a richer form of plan properties, specifying formulas over action subsets touched by the plan. We run comprehensive experiments on adapted IPC benchmarks, and find that the suggested analyses are reasonably feasible at the global level, and become significantly more effective at the local level.
    @inproceedings{Eiffer_AAAI2020,
    author = "Eiffer, Rebecca and Cashmore, Michael and Hoffmann, Joerg and Magazzeni, Daniele and Steinmetz, Marcel",
    title = "{A New Approach to Plan-Space Explanation: Analyzing Plan-Property Dependencies in Oversubscription Planning}",
    booktitle = "Proceedings of AAAI Conference on Artificial Intelligence (AAAI 2020)",
    year = "2020",
    publisher = "AAAI",
    address = "",
    pages = "",
    isbn = "",
    doi = ""
    }
  • Let’s Learn their Language? A Case for Planning with Automata-Network Languages from Model Checking. Joerg Hoffmann, Holger Hermanns, Michela Klauck, Marcel Steinmetz, Erez Karpas and Daniele Magazzeni; Proceedings of AAAI Conference on Artificial Intelligence (AAAI 2020), 2020.

    Abstract:
    It is widely known that AI planning and model checking are closely related. Compilations have been devised between various pairs of language fragments. What has barely been voiced yet, though, is the idea to let go of one’s own modeling language, and use one from the other area instead. We advocate that idea here – to use automata-network languages from model checking instead of PDDL – motivated by modeling difficulties relating to planning agents surrounded by exogenous agents in complex environments. One could, of course, address this by designing additional extended planning languages. But one can also leverage decades of work on modeling in the formal methods community, creating potential for deep synergy and integration with their techniques as a side effect. We believe there’s a case to be made for the latter, as one modeling alternative in planning among others.
    @inproceedings{Hoffmann_AAAI2020,
    author = "Hoffmann, Joerg and Hermanns, Holger and Klauck, Michela and Steinmetz, Marcel and Karpas, Erez and Magazzeni, Daniele",
    title = "{Let's Learn their Language? A Case for Planning with Automata-Network Languages from Model Checking}",
    booktitle = "Proceedings of AAAI Conference on Artificial Intelligence (AAAI 2020)",
    year = "2020",
    publisher = "AAAI",
    address = "",
    pages = "",
    isbn = "",
    doi = ""
    }
  • An Introduction to the Planning Domain Definition Language. Patrik Haslum, Nir Lipovetzky, Daniele Magazzeni and Christian Muise; Synthesis Lectures on Artificial Intelligence and Machine Learning, 2019.

    Abstract:
    Planning is the branch of Artificial Intelligence (AI) that seeks to automate reasoning about plans, most importantly the reasoning that goes into formulating a plan to achieve a given goal in a given situation. AI planning is model-based: a planning system takes as input a description (or model) of the initial situation, the actions available to change it, and the goal condition to output a plan composed of those actions that will accomplish the goal when executed from the initial situation. The Planning Domain Definition Language (PDDL) is a formal knowledge representation language designed to express planning models. Developed by the planning research community as a means of facilitating systems comparison, it has become a de-facto standard input language of many planning systems, although it is not the only modelling language for planning. Several variants of PDDL have emerged that capture planning problems of different natures and complexities, with a focus on deterministic problems. The purpose of this book is two-fold. First, we present a unified and current account of PDDL, covering the subsets of PDDL that express discrete, numeric, temporal, and hybrid planning. Second, we want to introduce readers to the art of modelling planning problems in this language, through educational examples that demonstrate how PDDL is used to model realistic planning problems. The book is intended for advanced students and researchers in AI who want to dive into the mechanics of AI planning, as well as those who want to be able to use AI planning systems without an in-depth explanation of the algorithms and implementation techniques they use.
    @book{PDDLbook2019,
    author = "Haslum, Patrik and Lipovetzky, Nir and Magazzeni, Daniele and Muise, Christian",
    title = "{An Introduction to the Planning Domain Definition Language}",
    booktitle = "Synthesis Lectures on Artificial Intelligence and Machine Learning",
    year = "2019",
    publisher = "Morgan \& Claypool Publishers 2019",
    address = "",
    pages = "",
    isbn = "",
    doi = "10.2200/S00900ED2V01Y201902AIM042"
    }
  • Automated Planning for Robotics. Erez Karpas and Daniele Magazzeni; Annual Review of Control, Robotics, and Autonomous Systems, 2020.

    Abstract:
    Modern robots are increasingly capable of performing “basic” activities such as localization, navigation, and motion planning. However, for a robot to be considered intelligent, we would like it to be able to automatically combine these capabilities in order to achieve a high-level goal. The field of automated planning (sometimes called AI planning) deals with automatically synthesizing plans that combine basic actions to achieve a high-level goal. In this article, we focus on the intersection of automated planning and robotics and discuss some of the challenges and tools available to employ automated planning in controlling robots. We review different types of planning formalisms and discuss their advantages and limitations, especially in the context of planning robot actions. We conclude with a brief guide aimed at helping roboticists choose the right planning model to endow a robot with planning capabilities.
    @article{Karpas_2020,
    author = "Karpas, Erez and Magazzeni, Daniele",
    title = "{Automated Planning for Robotics}",
    journal = "Annual Review of Control, Robotics, and Autonomous Systems",
    year = "2020",
    doi = "10.1146/annurev-control-082619-100135"
    }