Views: 0
The human hand is one of evolution's most extraordinary tools. It allows us to grip firmly, yet also perform movements of surprising delicacy. We can lift a heavy object, turn a tool, or pick up a tiny part almost without thinking.
The first hominids appeared approximately 6 million years ago. Since then, our hands have evolved into highly sophisticated mechanisms capable of combining movement, force, sensitivity, and coordination. Precisely for this reason, replicating a human hand in a robot poses an enormous challenge.
For years, the field of robotics has attempted to create hands that increasingly resemble our own. This makes sense: we live in a world designed for human hands and use tools intended for our five fingers. Yet, a fundamental question remains: does a robotic hand really need to be identical to a human hand?
Boston Dynamics appears to have answered "no." Its new hand for the humanoid robot Atlas uses four fingers—including the thumb—instead of the five found on a human hand. And this isn't simply a matter of removing a finger; the design was developed with the robot's actual tasks in mind.
The new hand features 13 degrees of freedom and tactile sensors in the fingers and palm. It is designed to perform manipulation tasks, reposition objects within the hand itself, and use industrial tools.
This decision is driven by interesting industrial logic. When manufacturing a prototype, complex solutions are permissible. However, when planning to manufacture tens of thousands of robots, every component must be reliable, repairable, simple to manufacture, and economically viable.
Adding a finger means more than just adding a finger. It entails incorporating actuators, joints, sensors, electronics, control systems, and additional weight and power consumption, while also increasing maintenance needs and the potential for failure.
Boston Dynamics explains that a fifth finger would have provided certain additional capabilities, but also three extra degrees of freedom and greater complexity. They concluded that the advantages did not outweigh the drawbacks for the intended task.
This highlights an engineering lesson that extends far beyond robotics: the best design is not necessarily the one with the most features, but the one that delivers the necessary functionality with the least possible complexity.
This is particularly important now that the Atlas robot is moving from the lab to real-world industrial applications. Boston Dynamics has begun deploying Atlas for industrial tasks at Hyundai production plants, where the robots are being trained in actual manufacturing environments.
The true test, therefore, is still underway. We will have to see how the hand performs after thousands of hours of work manipulating various parts, tools, and objects.
However, as a design concept, it strikes me as a very sensible solution: rather than simply copying nature, the goal is to learn from it and redesign elements to meet industry needs.
We will soon know if they got it right.
A video can be seen at: https://www.youtube.com/watch?v=4whgw2gLBS8
EN 
ES