AI, Robot: A year in review

As 2024 comes to a close, the robotics industry is still a long way from the all-in-one C-3PO droid of sci-fi daydreams. But thanks to AI, a robot can now put a T-shirt on a coat hanger—a seemingly simple task that shows just how far robotics research has come.

Researchers aren’t the only ones excited about these sorts of AI-driven robotics breakthroughs—investors are, too. Robotics startups are on track to raise nearly USD 7.5 billion by the end of 2024. And humanoid robots alone could become a USD 7 trillion market by 2050.

We spoke with Jose Favilla, global leader for Industry 4.0 at IBM, about the biggest trends in robotics this year—and what to look out for in 2025.

Manual dexterity in humanoids has come a long way this year. Google taught robots to tie shoelaces. Figure has one that can make coffee. And then there’s Tesla’s new Optimus hand, which has 22 degrees of freedom and can catch a tennis ball!

The strides made in robot dexterity this year are remarkable. They reflect a convergence of advanced hardware designs and sophisticated AI learning techniques. These developments not only enhance the functionality of robots but also pave the way for their integration into various sectors, ultimately transforming how we interact with technology in our daily lives, such as assisting in household tasks, caregiving or even performing complex surgeries.

The future of robot dexterity looks promising as ongoing research aims to close the gap between human and robotic manipulation capabilities. Key areas of focus include hardware development and advanced sensory, allowing the robot to interact with the environment more naturally, as well as enhanced AI learning algorithms, enabling the robots to perform a wider range of tasks.

In September, World Labs cofounder Fei-Fei Li suggested that vision-based AI, capable of truly seeing and understanding the physical world, is key to advancing robotics.

Fei-Fei Li's emphasis on vision-based AI underscores its transformative potential in robotics. By enabling machines to perceive and understand their environment as humans do, we are likely to witness a significant evolution in how robots operate and integrate into our daily lives.

Vision-based AI equips robots with the ability to perceive their surroundings, navigate through environments, recognize objects and interpret complex scenes. Robots will be able to analyze visual inputs in real time, enabling them to make informed decisions. This requires integration with other forms of inputs, such as sensors.

The advancements made this year indicate a promising future where intelligent robots enhance productivity and safety across numerous applications.

Conversely, researchers at MIT have argued that using large language models instead of expensive visual data is a better route to take.

While LLMs will play a crucial role in advancing robotics and particularly knowledge integration, there is a need for a holistic approach that embraces natural language processing to communicate via voice, visual-based AI and even other forms of foundation models that can interpret sensor data in the form of a time series. A holistic approach that embraces and integrates these technologies will likely yield the most significant advancements in robotic capabilities and applications in the coming years.

However, this approach requires sophisticated architectures that can effectively merge sensor data, visual perception and linguistic understanding. Developing such integrated systems poses technical challenges, but it also presents opportunities for innovation.

Humanoid robots have made major inroads in warehouse and automotive settings this year. Will they expand to other factory or business types in 2025?

By 2025, humanoid robots are set to expand their footprint beyond warehouses and automotive settings into diverse industries such as healthcare, retail, agriculture and domestic environments. The combination of mass production initiatives, technological advancements and a growing recognition of their potential benefits suggests that humanoid robots will play a crucial role in reshaping how we work and live. As these technologies mature, they are likely to foster new forms of human-robot collaboration that enhance productivity and improve quality of life across multiple sectors.

The global market for humanoid robots is projected to grow significantly due to advancements in AI and reductions in production costs. And with major players emerging from Asia, particularly China, the competitive landscape is becoming increasingly dynamic. This competition is likely to drive innovation and lower costs, making humanoid robots more accessible to various businesses.

You mentioned that there have been a lot of advancements in China. Are there any other regions we should keep an eye on?

The International Federation of Robotics (IFR) notes that China has become the fastest-growing robot market globally, due to its strategic government initiatives and investments aimed at establishing the country as a global leader in this field. The Chinese Ministry of Industry and Information Technology (MIIT) has set ambitious targets for the mass rollout of humanoid robots by 2025 and aims for advanced technological capabilities by 2027.

However, other countries—such as the United States, Japan, Europe and South Korea—are also poised to contribute to advancements in robotics. In the United States, companies like Tesla, with its Optimus robot, and Boston Dynamics, with Atlas, are pushing the boundaries of humanoid robotics. Japan is home to companies like SoftBank Robotics, the creator of Pepper, and Honda, with ASIMO. Countries like Germany and France are investing in robotic technologies.  South Korea is also emerging as a notable player in the robotics sector.

Speaking of Japan, this year Japanese scientists made a smiling robot with “living” skin.

The recent advancements in biohybrid robots, particularly the creation of a smiling robot with "living" skin in Japan, signify a groundbreaking shift in robotics that merges biological and synthetic components. This field is rapidly evolving and holds substantial promise for various applications across healthcare, environmental monitoring and beyond.

As research continues and ethical considerations are addressed, we can anticipate a future where biohybrid robots play a crucial role in enhancing human capabilities and addressing complex challenges across various domains.

The development of biohybrid robotics will require collaboration across multiple disciplines, including biology, engineering, materials science and ethics. This interdisciplinary approach will be crucial for addressing technical challenges and ensuring responsible innovation.

As biohybrid robotics advances, ethical questions arise regarding the treatment of these entities. Issues such as sentience, pain perception and the moral implications of creating life-like machines will need careful consideration as technology progresses.

There’s also been discussion about a single, general-purpose humanoid that can perform a wide range of tasks. NVIDIA CEO Jensen Huang, for example, announced Project GR00T, a foundation model for humanoid robots, earlier this year—and has said that someday our lives may be filled with “digital and physical agents.” In some respects, this seems like the biggest robotics news coming out of 2024.

The vision articulated by Jensen Huang regarding the future of humanoid robots and AI agents suggests a transformative shift in how we interact with technology. His predictions, particularly the introduction of GR00T and the concept of a general-purpose humanoid platform, point toward a future where digital and physical agents coexist and collaborate in daily life, presenting exciting possibilities for innovation and collaboration between humans and machines.

However, a cultural shift is required to foster the acceptance of humanoid robots in our everyday lives, requiring innovations that prioritize user-friendly interactions. As usually happens with the introduction of new technologies, we will see this happen gradually, starting with the early adopters and expanding to broader adoption over time.