Earlier this month at The Conversation, mechanical intelligence researcher Hamed Rajabi noted that impressive-looking robots come with serious built-in limitations:

The core issue is that humanoid robots tend to be designed around software that controls everything centrally. This “brain-first” approach results in physically unnatural machines. An athlete moves with grace and efficiency because their body is a symphony of compliant joints, flexible spines and spring-like tendons. A humanoid robot, by contrast, is a rigid assembly of metal and motors, connected by joints with limited degrees of freedom.

“Today’s humanoid robots look remarkable, but there’s a design flaw holding them back,” August 18, 2025

He offers examples:

Tesla’s Optimus, for instance, is smart enough to fold a t-shirt. Yet the demonstration actually reveals its physical weakness. A human can fold a t-shirt without really looking, using their sense of touch to feel the fabric and guide their movements.

Optimus, with its relatively rigid, sensor-poor hands, relies on its powerful vision and AI brain to meticulously plan every tiny motion. It would likely be defeated by a crumpled shirt on a messy bed, because its body lacks the physical intelligence to adapt to the unpredictable state of the real world.

Boston Dynamics’ new, all-electric Atlas is even more impressive, with a range of motion that seems almost alien. But what the viral acrobatics videos don’t show is what it can’t do. It could not walk confidently across a mossy rock, for instance, because its feet cannot feel the surface to conform to it. It could not push its way through a dense thicket of branches, because its body cannot yield and then spring back. “Design flaw holding them back”

This, he stresses, is a built-in limitation of the current robotics industry: “today’s top robotics firms are fundamentally software and AI companies, whose expertise lies in solving problems with computation. Their global supply chain is optimised to support this with high-precision motors, sensors and processors.”

In short, more sophisticated software will yield only limited further advances. He advocates a new approach, based on “advanced materials and biomechanics.”

The mechanical intelligence approach

Rajabi, who is Founder and Director of the Mechanical Intelligence (MI) Research Group at London South Bank University, sees mechanical intelligence as “exploring how intelligence can emerge from the complex interaction of material and structure, and how these principles can be formalised and then embedded directly into engineering systems.”

Can actual intelligence emerge from the “complex interaction of material and structure” that is not even alive? At the journal Advanced Science, Rajabi and a colleague offer,

Despite significant scientific advances in the past decades, most structures around us are static and ironically outdated from a technological perspective. Static structures have limited efficiency and durability and typically perform only a single task. Adaptive structures, in contrast, adjust to different conditions, tasks, and functions. They not only offer multi functionality but also enhanced efficiency and durability. Despite their obvious advantages over conventional structures, adaptive structures have only been limitedly used in everyday life applications. This is because adaptive structures often require sophisticated sensing, feedback, and controls, which make them costly, heavy, and complicated. To overcome this problem, here the concept of Mechanical Intelligence (MI) is introduced to promote the development of engineering systems that adapt to circumstances in a passive automatic way. MI will offer a new paradigm for designing structural components with superior capabilities. As adaptability has been rewarded throughout evolution, nature provides one of the richest sources of inspiration for developing adaptive structures. MI explores nature inspired mechanisms for automatic adaptability and translates them into a new generation of mechanically intelligent components. MI structures, presenting widely accessible bioinspired solutions for adaptability, will facilitate more inclusive and sustainable industrial development, reflective of Goal 9 of the 2030 Agenda for Sustainable Development.

Khaheshi A, Rajabi H. Mechanical Intelligence (MI): A Bioinspired Concept for Transforming Engineering Design. Adv Sci (Weinh). 2022 Nov;9(32):e2203783. doi: 10.1002/advs.202203783. Epub 2022 Sep 14. PMID: 36104206; PMCID: PMC9661836.

Mimicking evolution?

The basic idea seems to be to try to build a mechanical device similar to an animal:

The tendons in the leg of a running hare act like intelligent springs. They passively absorb shock when the foot hits the ground, only to release the energy to make its gait stable and efficient, without requiring so much effort from the muscles.

Think about the human hand. Its soft flesh has the passive intelligence to automatically conform to any object it holds. Our fingertips act like a smart lubricator, adjusting moisture to achieve the perfect level of friction for any given surface.

If these two features were incorporated into an Optimus hand, it would be able to hold objects with a fraction of the force and energy currently required. The skin itself would become the computer. “Design flaw holding them back”

It’s not clear where the “evolution” part comes in. A non-living robotic device did not evolve from anything and it is not going to evolve into anything else. In reality, designing and building it would require a great deal of human intelligence, research time, capital, etc. in an industry whose resources are, as Rajabi admits, entirely tied up in a different model.

If a machine that operated in as complex a way as, say, a dog, were built, wouldn’t each component need to operate like a cell? Perhaps not; perhaps the robot dog could make do with less complexity than a natural one. But how much less complexity before a function that mimics nature is compromised? And there may be an upper limit to the level of complexity that can be economically engineered, built, and maintained.

Everything in our universe reaches a ceiling eventually and it will be interesting if mechanical intelligence ends up being the ceiling for robotics.