LLMs Meet Robotics: The Rise of AI-Driven Machines in 2024 Tech giants are leveraging LLMs to make robots faster, smarter, and more cost-effective to deploy across industries
Opinions expressed by Entrepreneur contributors are their own.
You're reading Entrepreneur India, an international franchise of Entrepreneur Media.
In 2024, Large Language Models (LLMs), a subset of artificial intelligence (AI), like Google DeepMind's Gemini, OpenAI's ChatGPT, Anthropic's Claude models, Mistral AI's open-source models, and others have been dominating the tech space globally. Now, a trend of integrating LLMs in robotics is gaining traction. Nathan Benaich, founder and solo general partner at the early-stage AI investment firm Air Street Capital, highlighted this resurgence in the 8th annual 'State of AI' report. Tech giants are leveraging LLMs to make robots faster, smarter, and more cost-effective to deploy across industries.
DeepMind's AutoRT: Smarter Robotics with AI Integration
DeepMind's latest system, AutoRT, combines two AI models to enhance robot intelligence. One model allows the robot to understand its surroundings, while the other suggests tasks the robot can perform. Together, these models enable robots to quickly adapt to new environments and operate with minimal delays.
In addition, DeepMind introduced RT-Trajectory, a novel method for training robots through videos. By incorporating 2D sketches of robotic actions into training videos, robots can more easily learn and replicate tasks. The new SARA-RT method also optimizes robotic control systems, requiring less computing power while maintaining high performance. These innovations, along with Gemini 1.5 Pro, enable robots to respond more naturally to human instructions and function efficiently in real-world scenarios.
Hugging Face's Open Platform: Lowering Barriers to Robotics
Access to robotics has traditionally been challenging due to the lack of open datasets and tools. Hugging Face, a French-American company backed by Amazon, Nvidia, and Google, is changing the landscape with LeRobot, an open platform offering pre-trained models, datasets, and robotic demonstrations. This initiative makes it easier for enthusiasts and researchers to experiment, explore, and innovate in robotics, democratizing the field.
The state of LLMs and robotics at home
In 2023-24, several Indian companies made their foray into the LLM market. Notably, Bhavish Aggarwal, co-founder of Ola, launched Krutrim, while CoRover.ai introduced BharatGPT in collaboration with Bhashini under the National Language Translation Mission. Known for its conversational AI, CoRover provides virtual assistants to organizations like IRCTC and the Government of India, with a user base exceeding one billion.
On the robotics front, India's first indigenous humanoid robot, Mitra, was designed for smart human interaction. Meanwhile, another humanoid robot Manav holds the distinction of being the country's first 3D-printed robot. Additionally, Hyderabad-based H-Bots Robotics developed Robocop, an AI-powered police robot designed to assist in law enforcement, order management, and traffic control, and the list goes on.
However, when comparing India's robotics market to that of developed nations, a significant gap is evident. The Indian robotics market is projected to generate USD 446.70 million in revenue in 2024, while the U.S. market is expected to reach USD 9.42 billion in the same year. But, in recent years, India witnessed tremendous tech transformation and a lot of tech startups are emerging. In coming years, India might transform this gap.
"A number of research groups are aiming to bridge the gap between high-dimensional observation and low-dimensional action spaces in robot learning," said Benaich.
