As we seek to create robots that’re more ‘human’ who’s helping? AI

As robotics progresses towards creating humanoid robot helpers, our tendency is to create them in our own image. And our good buddy Artificial Intelligence (AI) is right by our side helping.

How soon can’t be said yet, but our adoption and affinity towards humanoid robot technology might depend on how much alike to us these bipedal machines are.

If you haven’t kept track of how far robots have come, you’ll find that while we were busy with AI’s super progress, robots have come way ahead. And they aren’t competing with AI, instead they’ve adopted AI. Robots are not only coming into focus as AI gets into the robotics industry, but AI is also helping create robots that seem more human.

Going Humanoid

Humanoid robots seem to be nearing the common consumer, looking at the rapid development in the area.

This week, Meta unveiled an AI robotics model called V-JEPA 2. The tech giant claims that V-JEPA 2, which is designed to understand movements of objects to enhance the technology of machines such as delivery robots and self-driving cars, has a better understanding of the physical world.

In March, Google launched a version of its AI model, Gemini that wows us not just digitally but phygitally. Gemini Robotics blends the power of large language models with spatial reasoning, meaning a user can tell a robotic arm to do something physically. The LLM filters the commands and then identifies the user’s intentions and then breaks them down into commands that the robot can carry out. 

Humanoid robots seem to be nearing the common consumer, looking at the rapid development in the area.

In another humanoid feat helped by AI, in February, the robotics company Figure showed off two humanoids helping each other put groceries away, according to The Robot Report.

In the world of robotics, ‘world models’ are trending within the AI community as researchers look beyond large language models. World models are systems that take inspiration from the logic of the physical world to create an internal simulation of reality, helping the AI to learn, plan, and even make decisions in a more human-like manner.

Make Robots that ‘Do’ Human

We humans are inclined to create robots that look and ‘do’ human.  

Japanese scientists have found a way to stick living skin tissue on robotic faces to make them ‘smile’, in a breakthrough that promises innovation in cosmetics and medicine. The result, though eerie, is an important step towards building more life-like robots, a researcher said.

We humans are inclined to create robots that look and ‘do’ human.

Recently, researchers at École polytechnique fédérale de Lausanne (EPFL) in Switzerland developed an Addams Family Thing like robot arm that can detach its hand to grab things while crawling around. The research team generated and refined the design using a genetic algorithm that relies on biological tricks like natural selection and evolution and the MuJoCo physics simulator to test the practicality of iterations. The algorithm and simulations helped the researchers decide the optimal position and number of articulated fingers needed, which turned out to be five, in a similar layout to human hands.

This could mean that the human body, which is what it is after a long evolutionary process, may be the optimum design for a robot.

How Robots Learn to be Human

General AI models for robotics are definitely helping humanoid robots learn faster than ever before.

To train better general-purpose robots, MIT researchers developed a versatile technique that combines a huge amount of heterogeneous data from many of sources into one system that can teach any robot a wide range of tasks.

Researchers are using unique ways to train today’s robots. Robots are now being trained with sound too. Researchers at Stanford University used sounds of chores like flipping a bagel in a pan, erasing a whiteboard, putting two Velcro strips together, and pouring dice out of a cup, to help robots learn better than just video. It worked too. By including sound, the robot’s ability to tell if the dice were in the cup rose from 27% to 94%.

Ironically, to make a humanoid robot more ‘human’, it has to learn not just the efficient capabilities of a human being but also its flaws. Mohammed Brueckner, a techie told MIT Tech Review, that to be more useful, robots need to become lazier.

“We, as humans, inherently understand energy conservation, focusing our efforts where they matter most. This “selective laziness,” prioritizing tasks and dynamically adjusting strategies based on context, could be the key to unlocking the next level of robotic utility. It’s about integrating seamlessly into our world, not just performing tasks in isolation,” — Mohammed Brueckner,  techie

“We, as humans, inherently understand energy conservation, focusing our efforts where they matter most. This “selective laziness,” prioritizing tasks and dynamically adjusting strategies based on context, could be the key to unlocking the next level of robotic utility. It’s about integrating seamlessly into our world, not just performing tasks in isolation,” he says.

Meanwhile, music researchers in Dresden debuted a three armed robot conductor with the intention of playing music that human conductors would find impossible to do. This is where the robot veers ahead of the human imitating it, while doing things faster and better.

It’s beginning to look like human households must soon make space for another addition, a robot helper. How soon can’t be said yet, but our adoption and affinity towards humanoid robot technology might depend on how much alike to us these bipedal machines are.