Research themes & projects

Robotics & Control

My team works on a wide range of robotics and control projects and is generally interested in how automation can be used to help humans in tasks that are either too dangerous, too time-consuming or require operators with high technical expertise.

Examples of past and current projects include:

• Radiation-aware navigation of autonomous robots (rover, quadruped) in nuclear environments, to survey unknown structures while minimising the dose of radiation absorbed by the robot;
• Automatic design of pneumatically-actuated soft graspers embedding impedance control without requiring any sensors or real-time feedback;
• Bespoke grippers for lab automation (e.g. vial grasping in Chemistry labs);
• Automation of vertical farms;
• Development of autonomous mobile robots for preventative road maintenance (via spin-out Robotiz3d Technology)

On a more fundamental level, we are interested in how optimality (e.g. minimum time or energy consumption) can be balanced against robustness to noise or model uncertainty in presence of nonlinear dynamical models.

Sensing technologies

Availability of high-quality data is crucial to allow us to monitor and control devices and physical processes. My team can develop bespoke sensors for a wide range of applications, and improve the performance of existing sensing platforms by controlling them and exploiting their natural dynamical behaviour.

Examples of past and current projects include:

• Nanosensors for mass and viscosity sensing, exploiting self-excitation to dramatically improve response time and signal-to-noise ratio;
• Integrated platforms for mechanobiology, to monitor evolution of cells under nanokicking and accelerate diagnosis of leukemia;
• bespoke road scanning sensor, using laser profilometry to detect road defects with mm accuracy while travelling up to 100km/h;

We are also interested in how control techniques can take full advantage of data-rich scenarios made available with modern sensing technology, including use of AI/ML within control strategies.

Advanced Manufacturing

Manufacturing is ubiquitous in everyone's life, as all the products we use needs to be manufactured in the first place. My team is working on how to advance manufacturing processes to make them more reliable, sustainable and tailored to individual needs of end-users.

Examples of past and current projects include:

• Real-time detection of defects forming in Laser Powder Bed Fusion (metallic 3D printing) based on data collected during the build process. This include both lightweight statistical methods for fast computation and more advanced supervised/semi-supervised machine learning approaches for improved detection capabilities;
• Robotic-assisted large scale 3D printing, with printing heads hosted on mobile robots capable of either climb on the part they are building or of creating a support structure to enlarge their working volume in real-time
• programmable matter approaches, where products are composed of a large number of robotic modules capable of algorithmically changing shape and mechanical properties of the product itself.

Biomechanics & Healthcare devices

Biological systems excel in making mechanics, nonlinear dynamics and control work seamlessly in presence of often inaccurate sensors and actuators (compared to their engineering counterparts). My team is using mathematical modelling, data analysis and control techniques to better understand how humans and animal can achieve this, so that we can develop better sensors, assistive technology, rehabilitation devices etc.

Examples of past and current projects include:

• Creation of sensors to monitor human movements. For example, development of a depth camera-based platform to promote independent rehabilitation and exercise of frail people
• Development of bespoke wearable sensors and AI-driven analytics to monitor sleep positions potentially related to the occurrence of musculoskeletal problems;
• Mathematical model for proprioceptive-driven crawling, to explain how larvae and earthworms may move via propagation of contraction waves along the body without any central pattern generator;
• Use of soft robotics for post-stroke upperlimb rehabilitation.