Enabling self-learning and self-adaptation for materials, not people
Birds, bees and trees have one important behaviour in common. In their respective collectives – their flocks, swarms and forests – each individual member has the ability to receive and exchange signals, adapting to changing environmental conditions and improving how they function without anyone directing them. Inspired by this complex behaviour, the EU-funded IMMENSE(opens in new window) project has set out to determine whether materials and structures can be endowed with processes that simulate it – in other words, whether they can become sentient.
On the path to sentience
A sentient material or structure must be able to send and receive signals, and interpret and compare them, and so be able to self-learn and self-adapt. But what does this mean in practice? “A sentient structure could, for example, sense vibrations and wave propagation within its elements, autonomously interpret and classify the type of signals detected, and be able to react by changing the stiffness of some of its elements to mitigate the consequences of vibrations caused by external agents,” explains structural engineer Alberto Corigliano of IMMENSE project coordinator Politecnico di Milano, Italy, in an article(opens in new window) posted on the ‘Open Access Government’ website. “A sentient material could be able to detect local defects in advance and activate mechanisms for the release of substances for self-repair.” To realise this ambition, IMMENSE is making use of solid and structural mechanics, fluid-structure interaction and multi-physics phenomena at the micro- and mesoscale, combined with microstructured smart materials. The project is studying and developing innovative bio-inspired sensors and combining them with smart materials and metamaterials that can optimise the intensity of received signals and channel them wherever needed. Learning and reactive behaviour will be achieved by using the complex dynamics of oscillator arrays combined with innovative hardware for signal classification and recognition. Recent developments in analogue computing make it possible to build compact, energy-efficient devices that can independently carry out basic calculations. In one study(opens in new window) carried out in 2025, the project looked at a smart way to design truss structures using generative rules and a decision-making algorithm called Monte Carlo Tree Search. Allowing the system to explore many possible designs efficiently and focus on the most promising ones, this approach could be better suited to complex construction problems. Another IMMENSE study(opens in new window) describes how a highly flexible inkjet-printed piezoelectric sensor can be created on a thin polyimide substrate. Tests reveal that the sensor keeps working reliably even when it is bent sharply and repeatedly, suggesting this new printing approach could lead to affordable, customisable sensors for next-generation electronics. IMMENSE (Inter materials and structures mechanoperception for self learning) still has several theoretical and technological hurdles to overcome before it reaches its goal. However, armed with innovative new sensors, machine learning algorithms, and new systems for signal interpretation and classification, it is striving to create sentient materials with promising applications in fields such as biomedicine, engineering and construction. The project ends in 2029. For more information, please see: IMMENSE project website(opens in new window)
