blog
February 25, 2026
On January 6–7, 2026, Analog Devices, Inc. (ADI) participated and sponsored the 7th UK Robotic Manipulation Workshop at the University of Edinburgh. This world-class gathering brings together researchers and industry leaders focused on one of robotics’ hardest challenges: enabling robots to manipulate objects with the same adaptability and precision as humans. Over two days, experts shared progress in tactile sensing, control strategies, simulation, and industrial applications—culminating in a keynote that reframed how we think about manipulation.
Why Robotic Manipulation Matters
Manipulation is central to robotics. Whether assembling products, assisting in healthcare, or handling everyday objects, robots need to interact with the physical world safely and effectively. While robots excel in structured environments, real-world settings are unpredictable. Achieving human-level dexterity requires advances in sensing, control, and learning.
This workshop explored how to bridge that gap, with discussions ranging from tactile sensing and whole-body awareness to learning-based approaches and industrial use cases.
Key Themes and Talks
1. The “Full Stack” Problem in Manipulation
Among the many insightful talks, Matei Ciocarlie (Columbia University)’s talk was one of the best. Prof. Ciocarlie described manipulation as a “full stack” problem—spanning hardware, sensing, planning, and learning. He explained why traditional planning and control approaches struggle with contact-rich tasks: the state space is vast, uncertainty is high, and real-time reasoning is difficult.
He highlighted why reinforcement learning (RL) has succeeded where classical methods falter. RL incorporates uncertainty during training and distills policies for fast inference, making it suitable for tasks like in-hand manipulation. However, RL faces its own challenge: exploration. His team uses simulation to brute-force exploration and combines it with imitation learning from teleoperation to guide robots through complex state spaces.
Key insight: Success in dexterous manipulation is about co-design combining planning, RL, tactile sensing, and hardware design. Ciocarlie’s work shows that tactile sensors improve policy performance, with the future lying in hybrid approaches that balance sensing, learning, and mechanical design.
2. Designing Robots for Human-Level Dexterity
Shan Luo (King’s College London) introduced vision-based tactile sensors and SimTac, a simulation framework for tactile sensing. His work reduces trial-and-error in hardware design and improves generalization across tasks, showing how simulation accelerates progress in dexterity.
Opinions expressed in the content posted here are the personal opinions of the original authors, and do not necessarily reflect those of Analog Devices, Inc (ADI) or its subsidiaries. The content is provided for informational purposes only and is not meant to be an endorsement or representation by ADI or any other party. This site may also provide links or references to non-ADI sites and resources. ADI makes no representations, warranties, or other commitments whatsoever about any non-ADI sites or third-party resources that may be referenced, accessible from, or linked to this site.
3. Whole-Body Awareness for Safer Interaction
Perla Maiolino (University of Oxford) presented multimodal sensing systems combining time-of-flight sensors and artificial skin. These enable robots to anticipate collisions and adapt in real time, making collaboration safer and more natural.
4. Standards for Manipulation Pipelines
Adam Norton (UMass Lowell) discussed the COMPARE ecosystem, which aims to standardize robot manipulation pipelines for reproducibility and interoperability. In a novel field like robotic manipulation, performance evaluation remains a challenge. COMPARE will make it easier to share components and benchmark performance.
5. Learning and Control for Complex Tasks
Talks from University of Leeds’ Mehmet Dogar and University of Edinburgh’s Marina Aoyama focused on physics-based predictions and simulation-driven learning for contact-rich tasks. As covered by Yuke Zhu’s data pyramid concept, robotics data is scarce, complex, and expensive to obtain. Combining large simulated datasets with smaller real-world samples helps bridge the sim-to-real gap.
6. Language Meets Manipulation
Anh Nguyen (University of Liverpool) introduced language-driven grasp detection, where robots interpret natural language instructions—such as “pick up the mug by the handle”—and translate them into grasp strategies. This approach makes human-robot interaction more intuitive.
7. From Tactile Sensing to Adaptive Manipulation
The University of Bristol’s Efi Psomopoulou (pictured) focused on what robots needed to reach human hand-like dexterity levels. One point that resonated with was her team’s use of reinforcement learning for in‑hand manipulation. Their research showed that tactile sensing helps policies generalize across different hand orientations and object configurations.
8. Actuation for Dexterity
Guggi Kofod (Pliantics) shared progress in dielectric elastomer actuators, which mimic biological performance. These actuators offer compliance and strength but face challenges in cost and reliability.
9. Industrial and Sustainability Applications
Talks from Mustafa Suphi Erden and Maria Koskinopoulou showcased real-world applications:
10. How humans actually hold everything
University of Stirling’s Arran Reader presentation offered a useful human‑centred perspective on manipulation. His work examined how people choose grip types, sequence actions, and coordinate both hands when handling multiple objects. His research provides guidance for robotic systems that need to decide which object to grasp first, when to switch grips, and how to manage transitions between tasks.
Why This Matters
Robotic manipulation is more than a technical challenge—it’s a gateway to broader automation in homes, factories, and healthcare. Progress in tactile sensing, dexterity, and learning-based control will make robots safer, more capable, and more collaborative. These advances will impact industries from manufacturing to medicine and help address global challenges like sustainability. As an enabler of many of the world’s leading robotics projects and products, ADI invests in many of the technologies robotics developers rely on, from sensors and IMUs to connectivity. Stay tuned for more on our investments and partnerships in that space.
Related content: