ADVERTISEMENTs
.png)
Researcher leads breakthrough in robot’s ability to perform complex tasks
Kumar’s approach integrates large language models (LLMs) with real-world simulations to test and refine action plans, ensuring robots operate within environmental constraints.
Nishanth Kumar / Image- MIT
Indian American PhD student Nishanth Kumar, a researcher at Massachusetts Institute of Technology's Computer Science and Artificial Intelligence Laboratory (CSAIL), has contributed to the development of a strategy enabling robots to safely perform open-ended tasks.
These tasks, which often involve ambiguous or multi-step instructions such as "clean the room" or "make breakfast," require robots to adapt to dynamic environments while understanding their physical constraints.
Kumar’s approach, "Planning for Robots via Code for Continuous Constraint Satisfaction" (PRoC3S), integrates large language models (LLMs) with real-world simulations to test and refine action plans, ensuring robots operate within environmental constraints.
"LLMs and classical robotics systems like task and motion planners can’t execute these kinds of tasks on their own, but together, their synergy makes open-ended problem-solving possible," said Kumar, co-author of the study with fellow PhD student Aidan Curtis.
PRoC3S uses vision models to evaluate a robot’s surroundings and integrates this information with LLMs to simulate tasks such as drawing shapes, sorting blocks, and placing objects accurately. Tested extensively in both digital and real-world settings, the method achieved an 80 percent success rate in simulations and performed reliably in real-world trials involving robotic arms.
"We’re creating a simulation on-the-fly of what’s around the robot, trying out many possible action plans," Kumar explained. "Vision models help us create a very realistic digital world that enables the robot to reason about feasible actions for each step of a long-horizon plan."
This innovation has potential applications in household robots, enabling them to tackle complex chores like meal preparation or room cleaning. Future research aims to incorporate advanced physics simulators and adapt PRoC3S for mobile robots.
Eric Rosen, an AI researcher unaffiliated with the study, noted the significance of the work: “This combination of planning-based and data-driven approaches may be key to developing robots capable of understanding and reliably performing a broader range of tasks than currently possible.”
The study, supported by several U.S. research agencies, was recently presented at the Conference on Robot Learning in Munich, Germany.
Kumar completed his high school education at the Indian Public School in Tamil Nadu, where he graduated as valedictorian with a perfect score in the International Baccalaureate program. He later earned a bachelor’s degree in computer engineering with honors from Brown University and is now pursuing a PhD in electrical engineering and computer science at MIT under professors Leslie Pack Kaelbling and Tomás Lozano-Pérez.
ADVERTISEMENT
ADVERTISEMENT
ADVERTISEMENT
E Paper
ADVERTISEMENT
.jpg)
Video
Comments
Start the conversation
Become a member of New India Abroad to start commenting.
Sign Up Now
Already have an account? Login