In the late 1990s, engineer Rich Walker recalls building one of Shadow Robot’s earliest robotic hands using simple materials like wood, springs, and rubber bands. At the time, much of it was improvised with whatever parts were available. Walker, now a director at the London-based robotics firm Shadow Robot Company, contrasts that with the company’s modern designs, which are far more advanced and precise.

Today, their robotic hands feature cylindrical forearm units containing electric motors called actuators. These motors pull on tendon-like cables that control each finger with high precision. In demonstrations, users wear finger sensors that translate human movements into robotic motion. Even beginners can quickly manipulate objects like blocks and cups thanks to the system’s smooth responsiveness.

Around 200 of these robotic hands are currently used worldwide, mainly in universities and research labs. Walker describes them as development platforms for studying dexterity, helping researchers understand what kinds of capabilities future robots will need before designing full-scale systems.

Experts say that building a capable robotic hand is one of the biggest hurdles in robotics. Bren Pierce, founder of Bristol-based start-up Kinisi, notes that the human hand remains extremely difficult to replicate. His company’s KR1 robots are already being tested in commercial environments, equipped with interchangeable tools such as heavy-duty grippers or suction devices depending on the task.

Pierce believes the long-term goal is still a single, versatile robotic hand that can handle almost any job. Kinisi has developed a three-finger prototype that performs well, but he admits the real challenge lies in making it durable, scalable, and affordable. At present, the prototype costs around £4,000, compared with £400 for simpler gripping tools used in deployment.

The difficulty of creating a human-like robotic hand has also been highlighted by Tesla CEO Elon Musk. Speaking at a technology summit in Los Angeles, he identified dexterous hands, advanced AI perception, and mass production as the three biggest challenges in humanoid robotics. Musk has claimed that future versions of Tesla’s humanoid robot, Optimus, will eventually feature highly advanced, human-level hand dexterity.

However, researchers remain cautious. Professor Nathan Lepora from University of Bristol argues that human-level robotic dexterity is still years away. While progress is real, he estimates it could take roughly a decade rather than just a few years. His work, supported through the UK government’s ARIA UK research initiative, focuses on tendon-driven robotic hands similar in concept to those developed by Shadow Robot.

Lepora believes these tendon-based systems may ultimately prove more practical and affordable as they become more refined. At the same time, he points to rapid progress in China, where companies are integrating compact motors directly into robotic joints instead of relying on tendon systems.

One example is Shanghai-based Wuji Technology, whose latest robotic hand includes four independently controlled joints per finger, enabling highly detailed movement. The company also integrates piezoelectric sensors that provide tactile feedback by converting pressure into electrical signals, giving the robot a basic sense of touch.

Despite these advances, durability remains a major concern. Start-up founder Pierce notes that while laboratory sensors can perform well, they often fail after months rather than years. For real-world industrial use, robots must operate reliably for far longer periods—potentially up to a decade.

Still, researchers are optimistic. Lepora says tactile sensing was once seen as something perpetually “ten years away,” but large-scale investment in humanoid robotics is accelerating progress. As funding increases and companies compete globally, breakthroughs in robotic hands may finally be moving closer to reality.