Our goal is to understand the nature of intelligent behavior in the physical world, through the study of human intelligence and the design and implementation of intelligent robots.
We bring expertise in perception, sensing, language, learning and planning, and have the objective of integrating these disciplines to make physical agents with human-like intelligence.
Diffusion models have proven effective as a robotic policy representation class—but thus far have mainly been applied in the imitation learning setting for relatively simple tasks. What does it take to scale the approach to long-horizon tasks that also require very fine control?
To simulate more training videos for robots, “GenSim2” uses multimodal and reasoning LLMs that can see and ponder about real-world data. The framework then helps teach a robot to imitate these synthetic data to better learn household and manufacturing tasks.
“Co-LLM” algorithm helps a general-purpose AI model collaborate with an expert large language model by combining the best parts of both answers, leading to more factual responses.
New algorithm helps robots practice skills like sweeping and placing objects, potentially helping them improve at important tasks in houses, hospitals, and factories.
CSAIL researchers introduce a novel approach allowing robots to be trained in simulations of scanned home environments, paving the way for customized household automation accessible to anyone.
Neural network controllers provide complex robots with stability guarantees, paving the way for the safer deployment of autonomous vehicles and industrial machines.
New CSAIL research highlights how LLMs excel in familiar scenarios but struggle in novel ones, questioning their true reasoning abilities versus reliance on memorization.
CSAIL researchers’ soft robotic system can handle a continuous stream of unpredictable objects on a conveyor belt, potentially guiding robots with carefully packing groceries and recycled materials.
MIT CSAIL’s frugal deep-learning model infers the hidden physical properties of objects, then adapts to find the most stable grasps for robots in unstructured environments like homes and fulfillment centers.
MIT CSAIL researchers enhance robotic precision with sophisticated tactile sensors in the palm and agile fingers, setting the stage for improvements in human-robot interaction and prosthetic technology
MIT undergraduate and CSAIL member Srinath Mahankali, along with students Ben Lou and Kenta Suzuki, are honored for their academic excellence, with research exploring math and physics.
Three neurosymbolic methods help language models find better abstractions within natural language, then use those representations to execute complex tasks.
MIT CSAIL postdoc Nauman Dawalatabad explores ethical considerations, challenges in spear-phishing defense, and the optimistic future of AI-created voices across various sectors.
Adaptive smart glove from MIT CSAIL researchers can send tactile feedback to teach users new skills, guide robots with more precise manipulation, and help train surgeons and pilots.
MIT CSAIL's innovative AI system melds simulations and physical testing to forge materials with newfound durability and flexibility for diverse engineering uses.
A multimodal system uses models trained on language, vision, and action data to help robots develop and execute plans for household, construction, and manufacturing tasks.
GenSim uses the code within LLMs to automatically generate new robotic behaviors and outline each step within long-horizon goals to assist with manufacturing, household, and logistical tasks.
By blending 2D images with foundation models to build 3D feature fields, a new MIT method helps robots understand and manipulate nearby objects with open-ended language prompts.
CSAIL researchers helped design an artificial intelligence algorithm that generates novel designs for robotic locomotion at impressive speeds while running on a lightweight personal computer.
Designed to ensure safer skies, “Air-Guardian” blends human intuition with machine precision, creating a more symbiotic relationship between pilot and aircraft.
PIGINet leverages machine learning to streamline and enhance household robots' task and motion planning, by assessing and filtering feasible solutions in complex environments.
Daniela Rus, CSAIL Director and MIT professor, and a team of researchers from the Singapore-MIT Alliance for Research and Technology and the National University of Singapore earned an “Outstanding Navigation Paper” at the ICRA 2023 for their work on a method that evaluates the uncertainty measurements of the 3D LiDAR technology in autonomous vehicles.
Researchers create new simulation tool for robots to manipulate complex fluids in a step toward helping robots more effortlessly assist with daily tasks.
Through the Multidisciplinary University Research Initiative, the US Department of Defense supports research projects in areas of critical importance to national defense.
The CSAIL scientist pushes forward natural language processing research by creating state-of-the-art machine learning models and investigating how language can enhance other types of artificial intelligence.
MIT researchers exhibit a new advancement in autonomous drone navigation, using brain-inspired liquid neural networks that excel in out-of-distribution scenarios.
The system is cost-effective, scalable, and provides high-resolution with an “intelligent carpet” touch sensor, opening doors to VR gaming and healthcare applications.
"DribbleBot" can maneuver a soccer ball on landscapes such as sand, gravel, mud, and snow, using reinforcement learning to adapt to varying ball dynamics.
MIT CSAIL researchers solve a differential equation behind the interaction of two neurons through synapses to unlock a new type of speedy and efficient AI algorithm.
Yilun Du, a PhD student and MIT CSAIL affiliate, discusses the potential applications of generative art beyond the explosion of images that put the web into creative hysterics.
MIT scientists unveil the first open-source simulation engine capable of constructing realistic environments for deployable training and testing of autonomous vehicles.
A new technique can safely guide an autonomous robot without knowledge of its environmental conditions or the size, shape, or location of obstacles it might encounter.
“Evolution Gym” is a large-scale benchmark for co-optimizing the design and control of soft robots that takes inspiration from nature and evolutionary processes
If you see a self-driving car out in the wild, you might notice a giant spinning cylinder on top of its roof. That’s a lidar sensor, and it works by sending out pulses of infrared light and measuring the time it takes for them to bounce off objects. This creates a map of 3D points that serve as a snapshot of the car’s surroundings.
This week it was announced that MIT professor and CSAIL principal investigator Tomas Lozano-Perez has been awarded the 2021 IEEE Robotics and Automation Award for his “foundational contributions to robot motion planning and visionary leadership in the field.”
Using UVC light, the team's system can disinfect a warehouse floor in half an hour - and could one day be used in grocery stores, schools and other spaces
In a pair of papers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), two teams enable better sense and perception for soft robotic grippers.
Most robots are programmed using one of two methods: learning from demonstration, in which they watch a task being done and then replicate it, or via motion-planning techniques such as optimization or sampling, which require a programmer to explicitly specify a task’s goals and constraints.
