The discovery and synthesis of functional materials underpin advances in energy conversion, catalysis, and sustainable technologies. However, identifying optimal synthesis protocols is still dominated by trial-and-error experimentation. Nanoparticles are particularly challenging to synthetically control due to the large number of tunable parameters and the requirement for advanced structural characterisation. At the same time, their large surface areas and atomically defined arrangements make them highly promising candidates for next-generation energy technologies.

In this talk, I will present recent progress towards self-driving laboratories (SDLs) for inorganic nanoparticle synthesis, where robotic experimentation, synchrotron-based total scattering, and machine learning (ML) are integrated into closed-loop discovery platforms.

I will describe proof-of-concept SDL experiments in which robotic synthesis was optimised by a Bayesian optimisation algorithm directly informed by experimental scattering data. Within four days of synchrotron beamtime, the system autonomously developed synthesis protocols for target ~5 nm decahedral and ~10 nm face-centred cubic gold nanomaterials. By coupling our SDL with ML-driven interatomic potentials, we foresee that one can first identify promising material candidates in silico before they are realised synthetically. By bridging chemistry, automation, and AI, this research aims to transform materials discovery; moving from intuition-driven exploration to autonomous, knowledge-guided design at scale.

The possibilities for discovering new materials are boundless, yet physical resources are limited and the demand for functional materials to address climate and environmental challenges is more urgent than ever. In recent years, artificial intelligence (AI) and robotics have emerged as a transformative approach to accelerate scientific discovery. By coupling automated experiments with AI-driven decision-making, self-driving laboratories promise to reduce the materials innovation cycle from years to months. In this talk,I will highlight three complementary approaches to building autonomous research platforms spanning organic synthesis, halide perovskites, and metal-organic frameworks. First, I will discuss the advantages and limitations of all-in-one liquid-handling systems for high-throughput exploration of synthesis conditions. Second, I will introduce low-cost, DIY robots built from open-source hardware to democratize access to laboratory automation and adapt to evolving research needs from ex situ to in situ measurements. Third, I will present modular thin-film deposition systems designed to empower scientific creativity through human-in-the-loop autonomy, exemplified by the optimization of solution-processed semiconductors. Together, these strategies demonstrate how automation and AI can augment human expertise, accelerating the synthesis and characterisation of diverse classes of advanced materials and opening new pathways toward next-generation energy technologies. I will conclude by discussing the open challenges of interfacing AI with real-world scientific experiments and the emerging opportunities the field presents, envisioning a future of seamless human-AI-robot collaboration for materials discovery. 

Next-generation technologies in energy storage, molecular separations, and optoelectronics depend critically on how molecules assemble in the solid state. Small changes in packing can switch porosity on or off, alter diffusion pathways, or dramatically change charge transport.  Being able to predict and control the supramolecular assembly of materials remains one of the central challenges in molecular materials design. Yet the fundamental question of determining what phase a molecule will form – whether crystalline, disordered, or amorphous – and the atomic structure it adopts remains a major unsolved problem. In this talk, I will present an “assembly-aware” digital framework for predicting the solid-state structure of molecular materials that couples atomistic calculations on molecular dimers with coarse-grained statistical mechanical models, separating the influence of molecular shape and intermolecular interactions on supramolecular assembly. These models have the potential to act as fast, physics-informed surrogates inside high-throughput or self-driving workflows, guiding synthesis towards regions of chemical space that are both structurally and functionally promising.

Thanks to their versatility, rich capabilities, and potential to interact naturally with humans, humanoid robots are emerging as a solution for a wide range of tasks across domains spanning manufacturing, service robotics, and healthcare. 

In this talk, I will provide an overview of my group’s work on developing humanoid robots that are capable of acting autonomously in human environments. Specifically, I will discuss advances in visual recognition, tactile perception, and object manipulation.I will focus on the challenges of building systems that can learn autonomously from interaction with the real world while integrating knowledge acquired both off-line and on-line. I will then illustrate how these methods can be integrated in complete systems and deployed across different applications, with examples from service robotics and human–robot collaboration. The talk will highlight how humanoid embodiment and recent advances in AI support the development of robots capable of flexible and effective interaction in everyday scenarios.

With the maturation of the perovskite field, which has expanded dramatically over the past two decades, the challenge of identifying new materials has become increasingly demanding. Since their discovery in the 19th century, when naturally formed perovskites attracted limited interest, synthetic perovskites have gained remarkable attention in the mid-1990s. They now span a broad range of properties, including tunable bandgap,[1] defect tolerance,[2] mechanical flexibility[3] and strong light absorption.[4] 

Among these materials, layered perovskites[5–7] stand out for their hybrid organic-inorganic architecture, which enable fine control over emission,[8, 9] mechanical robustness[10] and ambient stability.[11] As the number of studied compositions and structural variants continues to grow, the search for innovative materials becomes increasingly complex. 

To address this challenge we introduce an automated synthesis framework built upon efficient low-temperature synthesis[12–14] previously developed in our lab. We coupled this strategy with the creation of digital twins for real-time experiments and data collection. This combined strategy enables faster feedback loops and enhanced throughput compared to traditional trial and error workflows.

Accelerated materials discovery is critical for addressing global challenges. However, developing new laboratory workflows still relies heavily on real-world experimental trials, which limits scalability due to the need for numerous physical make-and-test iterations. This talk presents Matterix, a multi-scale, GPU-accelerated robotic simulation framework designed to create high-fidelity digital twins of chemistry labs. This digital twin simulates robotic physical manipulation, powder and liquid dynamics, device functionalities, heat transfer, and basic chemical reaction kinetics. These capabilities are enabled by integrating realistic physics simulation and photorealistic rendering with a modular GPU-accelerated semantics engine, which models logical states and continuous behaviors to simulate chemistry workflows across different levels of abstraction. Matterix streamlines the creation of digital-twin environments through open-source asset libraries and interfaces, while enabling flexible workflow design via hierarchical plan definition and a modular skill library that incorporates learning-based methods. This approach enables sim-to-real transfer in robotic chemistry setups, reducing reliance on costly real-world experiments and allowing hypothetical automated workflows to be tested entirely in silico.

This talk introduces the concept of Cognitive Robotics, i.e. the field that combines insights and methods from AI, as well as cognitive and biological sciences, to robotics (cf. Cangelosi & Asada 2022 for book open access). This is a highly interdisciplinary approach that sees AI computer scientists and roboticists collaborating closely with psychologists and neuroscientists. We will use the case study of language learning to demonstrate this highly interdisciplinary field, presenting developmental psychology studies on children’s language acquisition and robots’ experiment on language learning.

Growing theoretical and experimental psychology research on action and language processing and on number learning and gestures in children and adults clearly demonstrates the role of embodiment in cognition and language processing. In psychology and neuroscience, this evidence constitutes the basis of embodied cognition, also known as grounded cognition. In robotics and AI, these studies have important implications for the design of linguistic capabilities, in particular language understanding, in robots and machines for human-robot collaboration. This focus on language acquisition and development uses Developmental Robotics methods, as part of the wider Cognitive Robotics approach. During the talk we will present examples of developmental robotics models and experimental results with the baby robot iCub and with the Pepper robot. One study focuses on the embodiment biases in early word acquisition and grammar learning. The same developmental robotics method is used for experiments on pointing gestures and finger counting to allow robots to learning abstract concepts such as numbers. We will then present a novel developmental robotics model, and human-robot interaction experiments, on Theory of Mind and its relationship to trust. This considers both people’s Theory of Mind of robots’ capabilities, and robot’s own ‘Artificial Theory of Mind’ of people’s intention. Results show that trust and collaboration is enhanced when we can understand the intention of the other agents and when robots can explain to people their decision making strategies.  

The implications for the use of such cognitive robotics approaches for embodied cognition in AI and cognitive sciences, and for robot companion applications will also be discussed. The talk will also consider philosophy of science issues on embodiment and on machine’s understanding of language, the ethical issues of trustworthy AI and robots, and the limits of current big-data large language models.