Robot Dexterity

Three-dimensional force and temperature sensing skins

Team: Guolin Yun + Zhuo Chen, University of Cambridge

Tawfique + team are developing a next-generation electronic skin (e-skin) for robots, enabling real-time, high-resolution 3D force and temperature sensing. Inspired by human touch, their multiscale structured design allows robots to detect force magnitude and direction, sliding, texture, surface stiffness, and temperature, achieving unprecedented dexterity and precision. By advancing robotic perception, they aim to enhance human-robot collaboration, unlocking new possibilities in surgical robotics, agriculture, precision manufacturing, and AI-driven automation.

R-ADA: A Rational Automated Design Agent

Team: Sam Witty, Eli Bingham, Michelangelo Naim, Tim Cooijmans + Cambridge Yang, Basis Research Institute

Zenna + team aim to develop AI technology for robotic design that mimics the workflows of expert robot designers and engineers. Their approach iterates between language model-driven program synthesis of robot design programs with simulation-based feedback on the quality of synthesised designs. Their work will substantially improve the speed and quality of robot design workflows, feeding into their core mission of developing general-purpose AI reasoning technology, then using that technology to solve critical societal challenges.

Computer-Aided Invention Of Robot Manipulators Across Domains

Team: Ayna Arora, Allan Zhao + Han Wang, MorphoAI

MorphoAI is developing a new paradigm of computer-aided invention to accelerate the design and deployment of next-gen cyber-physical systems. Current design cycles for robots, medical devices, and heavy machinery --- not to mention wholesale assembly lines --- can take weeks, months, or years. But no matter the complexity of the problem, iteration cycles take up significant time and effort. This team is developing simulation-driven AI tools that allow engineers to (semi-)automatically design and optimise new robotic hardware from high-level specifications, reducing trial-and-error. MorphoAI’s mission is to improve both how engineers design and raise the bar of what engineers can design.

A New Analytical Framework for Developing Dexterous Soft Robotic Manipulators

Patrick Keogh, Chris Bowen, Cangxiong Chen, University of Bath | Rika Antonova, University of Cambridge | Andrew Spielberg, MorphoAI

Min’s team aims to create a new analytical framework involving new model- and learning-based approaches and their integration to open new horizons in understanding the behaviours, fundamental characteristics, dynamics, and internal physical interactions of soft manipulators. This new framework focuses on deployable capability and versatility in describing and modelling dexterous manipulators. This work will lead to the creation and exploitation of next-gen intelligent and dexterous soft manipulators in manufacturing, healthcare, assistance, exploration, and rescue, allowing innovation and growth to maximise their economic and societal benefits.

SILA Miniaturisation (SAMI)

Team: Jill Burnett + Craig Fletcher, Wavedrives Ltd | Lloyd Ash + James Madge, Electrified Automation Ltd | Jeff Graham + Will Stanley, Xor Software Ltd

Motivated by experience building humanoid robots and market demand, WaveDrives has developed and patented the Sarcomere Inspired Linear Actuation (SILA), a novel actuation technology designed to enable human-quality robot movement. Now, Graham + team will miniaturise the SILA technology, while maintaining its efficiency and power density – to provide the muscle-like actuation needed for dexterous robotic manipulators at human scale. Partners at Electrified Automation and Xor Software will bring specialist motor design and electronics development expertise to the project for a cross-functional effort.

DEXTER – Simulation for Dexterous Manipulation

Team: Thomas Pak, Augustinas Simkus, Bence Rochlitz + Ori Cohen, Vsim

Vsim will create a set of advanced simulation features to accurately simulate tensile sensors, deformable objects, and tendon systems. The team will also provide a Machine learning framework and tools to enable other ARIA Creators to create policies to train robots and refine designs. Their aim is to offer high-performance, accurate simulation technologies coupled with extensive tools to enable other Creators to test, validate, and use AI to optimise designs and control policies. Ultimately, this will act as a fundamental building block for leveraging embodied AI, reducing costs, accelerating R&D, and driving innovation.

Scalable Tactile Sensing Based On Electrical Impedance Tomography For Dexterous Multi-tool Use

Team: George Thuruthel, University College London

Fumiya + team are developing a technological framework for the world's most flexible and scalable tactile perception, using Electrical Impedance Tomography (EIT) for dexterous robot manipulation. They aim to develop a design and manufacturing method for tactile sensor systems that allows well-balanced performances between all major metrics, e.g. high sensory density, large sensory area coverage, high force sensitivity, mechanical robustness, and faster reaction time and learning. They’ll work to achieve task-level metrics and benchmarks, including real-time closed-loop robot control for force-sensitive tool-use manipulation tasks such as screwdrivers, scissors, pliers, and chopsticks.

Realising the Future of Dexterous Robotics with Flexible Electrohydraulic Artificial Muscles

Team: Efi Psomopoulou, University of Bristol

Nicholas and Efi aim to develop high-performance, soft artificial muscles called HASEL actuators, which provide transmission-free linear actuation to enable adaptable and cost-effective motion solutions. This project will develop new materials and architectures for HASELs to achieve a step-change in capabilities and realise the holistic performance and versatility required for dexterous manipulators. The team will also explore strategies for integration and control of HASEL actuators to accelerate their incorporation into robotic manipulators.

Synthetic Muscles for Robotic Dexterity

Rodrigo will develop high-performance synthetic muscles that mirror human muscles in strength, speed, and agility. Powered by cutting-edge dielectric elastomer microfibers, Elysium’s muscles enable unparalleled dexterity – enough to build a fully capable robotic hand with 20 degrees of freedom – all while being lightweight, energy-efficient, and cost-effective. Current robotic systems can be clumsy, power-hungry, and prohibitively expensive, stalling progress towards advanced agility. Rodrigo’s vision is to break these barriers and usher in an era where robots handle dangerous and difficult tasks across sectors.

Romex: Robot Muscles as Dexterous Linear Actuators

Team: Rasmus Beck + Reza Nikbakht, Pliantics

Guggi and team will build soft linear actuators, which are the ‘artificial muscles’ that robots need to interact physically with the world. Compared to the servos used today, this team’s actuators will be soft, flexible, lightweight and more efficient. Servo actuators for robotics can be stiff, rigid, heavy, inefficient, and must be managed with careful software programming to become safe for people to interact with. The team’s vision is for their actuators to enable stronger, safer, and more bio-aligned dexterous solutions in robotics, entertainment, automation, and healthcare.

The ARTHUR Hand

The ARTHUR hand is a bio-inspired robotic hand. It features soft, deformable contact surfaces and rich tactile sensing, in conjunction with an innovative hierarchical reinforcement learning stack. Udayan’s goal is to build prototypes in the UK, delivering a dual-armed system with a range of dextrous manipulation capabilities that far exceed what’s possible today, and commercialising results with a focus on applications in manufacturing.

Printed electronics components – a new toolbox for the dexterous robot

Team: DZP Technologies

Zlatka will develop novel printed electronic components, creating a toolbox for building dexterous robot manipulators, which are impossible to produce with existing rigid electronics. This requires innovative materials solutions, as well as new scientific models to analyse, simulate, and predict the electrical and electronic properties of the new components. Her vision is to transform the robots of tomorrow by introducing new materials and novel forms of electronics, helping to solve productivity challenges in manufacturing, agriculture, sustainability, and ageing populations.

Sangtera Joint Actuator

Sangtera is developing a next-generation robotic actuator by improving the size and reliability of microhydraulic actuators. The human hand’s 27 degrees of freedom allow it to perform precise, complex tasks, but existing robotic hands fall short due to bulky, inefficient actuators. Sangtera’s actuators, powered by surface tension, offer torque densities hundreds of times greater than traditional options, making gearless and compact designs possible. These actuators, having the size of finger joints, will enable robotic hands with bio-aligned dexterity. The team’s vision is to continuously improve the performance, compactness, and cost-effectiveness of these actuators, to transform robotic manipulation across industries.

Pneumatic Overbraided Actuating / Manipulator Components

David Newsam, Mike Newsam, + Richard Taylor, Stellar Advanced Concepts Ltd | Philipp Thies, Chris Edwards, Halim Alwi, + Faryal Khalid, University of Exeter | Sebastian Brown, Peninsula Medical Technologies

This project will prototype a next-generation pneumatic artificial muscle (PAM). Air muscles offer numerous benefits for achieving dexterous robots that are safer for interaction with humans and delicate objects, but further progress is needed. Michael and team will develop a novel stacked arrangement that offers high-strength precision grip, positional precision, and force feedback, beyond contemporary soft robotic muscles. This unique approach uses contact forces imparted along overbraided sleeves to allow precise control through internal pressure variations. The team’s mission is to design transformative adaptive structures for next-generation robotic systems and autonomous vehicles.

MagTecSkin: Novel Tactile Sensitive Electronic Skin based on Magnetic Technology

Team: Susana Cardoso, INESC Microsystems and Nanotechnologies | Emiliano Bilotti, Imperial College London | Stefan Escaida, Universidad de O'Higgins

Tactile sensing capabilities are crucial for manual dexterity, yet remain beyond the reach of today’s robots. While recently developed robotic skins can measure contact forces accurately, they cannot bend or stretch, and therefore they cannot cover complex robot parts, such as finger joints or deformable links. Lorenzo and team will develop an innovative skin based on magnetic technology that can measure 3D contact forces on multiple contact points, as well as bend and stretch. This will unlock full-cover articulated and soft robots, which will ultimately lead to vastly advanced robot dexterity in manufacturing, logistics, agriculture, healthcare, and beyond.

DexComp

Team: Hayden Cole, Simon Groves, Michael Elkington, Alex Smith, National Composites Centre

This team will develop a supply chain of UK-developed industrial scale robotic solutions that can collaborate, emulate, or even improve the hand tasks and motions of composite experts. To achieve future market demand and maintain competitive manufacturing advantage, the UK needs to scale the rate of composite product production. However, the composite industry is constrained by a finite number of expert composite technicians who need to manually layup and form complex parts to stringent quality controls and accuracy. This constraint provides a unique opportunity for dexterous robotic solutions, which Marc and team are seeking to deliver.

Variably Constrained Drives for Muscle-like Electrical Actuators

Team: Etienne Hocquard + Simon Jones, Createc | Darren Ball, Fortis | Lewis Meredith, EM Simulations

Matt and team are re-imagining the design of electric actuators for robotics to obtain physical attributes that more closely resemble biological muscle. They’ve been using robot arms for robot manipulation for years, but are consistently held back by the limits of gearboxes in the arms that have them, or in the low strength-to-weight ratio of arms that don’t. Their goal is to unlock bio-aligned strength and dexterity by making an actuator that can be light, fast, and compliant when required, yet also be slow, strong and stiff when appropriate – and be efficient in both modes.