Rodney Brooks, a prominent figure in robotics and co-founder of iRobot, has expressed skepticism about the current trajectory of humanoid robots. Brooks contributes decades of experience from the Massachusetts Institute of Technology (MIT). He’s questioning the premise behind what makes these popular approaches, the ones that companies like Tesla and Figure are taking so far, effective. As Brooks explained, the industry is plagued by major challenges that need to be conquered in order to achieve the promise of humanoid robots.
One area where Brooks has repeatedly raised alarms is on the current state of the art approach to teaching robots complicated tasks. This is what he says makes the current strategy of training robots by letting them watch videos of humans doing things such a bad idea. Rather, he thinks a complementary approach is needed, one that’s much more creative and flexible to advance the state of robotic dexterity. Indeed, the skepticism often deepens whenever these investments or technological leaps into humanoid robotics happen. Brooks contends that all of these advancements are missing operational approaches and real-world practices.
Looking ahead, Brooks predicts that successful humanoid robots will deviate significantly from their human-like appearances within the next 15 years. He pictures a day when these robots glided along easily and quickly on smart wheels. Equipped with dozens of arms and multitasking supersonic sensors, they’ll touch and feel their environments in ways we can’t even imagine. This shift from the traditional human figure to more functional designs is based on needs, not canons.
Brooks highlights a crucial limitation in current robotic designs: the human hand’s complexity. For one, as he points out, human hands possess close to 17,000 specialized touch receptors. These receptors allow for unparalleled dexterity and precision. On the other hand—to quote his article directly—he claims there’s no robot in existence that can do anything close to this. They all moved forward, but touch data has been stagnant,” he stresses. This last category is particularly important for advancing the state of the art in robotic manipulation.
Brooks points out that a robot twice the size of today’s models would generate eight times the destructive energy. This boost is a result of its much larger mass. This raises important questions regarding safety in robot design. It underscores the importance of energy efficiency, particularly as companies race to build bigger and better humanoid robots.
Brooks, of course, is not denying the obvious success machine learning has had in fields like speech recognition and image processing. He’s quick to stress that these advances rest upon decades of foundational technology. He claims that similar breakthroughs in robotics are only possible with a greater focus on the data behind tactile interactions. We have to get through a second, big barrier—the complete absence of a tradition for capturing touch data. This is a prerequisite to create genuinely impactful humanoid robots.
As the robotics industry matures, Brooks’ warnings are a necessary wake-up call to the challenges still facing us and ahead. His skepticism regarding current methodologies and predictions about the future direction of humanoid robots underscore the need for innovative thinking in robotic design and functionality.