TRAILS AI Summer Academy at the University of Maryland
TRAILS AI Summer Academy at the University of Maryland
Embedded Files
Summer 2024 Research Presentations
Summer 2024 Research Presentations
For more pictures of how much fun camp was this summer, click here!
TRAILS AI SUMMER 2024
A two-week summer camp, exposing students to personal growth, educational, and hands-on experiences in artificial intelligence presented by faculty, guest lecturers, and University of Maryland students.
Meet our Participants
Project Groups
Project Groups
When moving in a crowd, we’ve figured out how to stay close to the people in our group and move away from those who are not. Our behavior depends on the type of interaction. We can also infer others' objectives by implicit interaction with each other through motion. Robots, on the other hand, rely on explicit communication or instructions to avoid collisions or getting stuck while moving toward a goal. The ability to model these interactions can help robots predict the uncertain behavior of pedestrians in the absence of explicit communication.
Recent years have seen the tremendous successes of machine learning, especially reinforcement learning (RL), where an agent makes sequential decisions by interacting with the environment via trial and error. Prominent and high-profiled examples include AlphaGo, an intelligent Go game-playing agent that has beaten the human champions, autonomous driving, and more recently the training of large-language models such as ChatGPT.
The use of basic imitation and reinforcement learning algorithms within manipulation environments such as MetaWorld and RoboSuite. Our focus is on long-horizon tasks, designing algorithms that not only perform isolated tasks but also seamlessly integrate these tasks into coherent long-term actions. This project involves hands-on implementation on actual robotic arms, encompassing data collection, algorithm development, and execution both in simulation and on real-world hardware.
People with disabilities (PwD) are the largest minority group in the world. For PwD, AI carries much promise and can create a more accessible world by (i) facilitating the design of novel technologies (ii) creating personalized technology experiences and (iii) scaling technology deployments. Yet, this promise is undermined by emergent concerns about the risks AI are likely to pose to PwD, such as threats to individual privacy, bias and discrimination.
With the growing use of AI in high-stakes applications such as hiring, lending, etc., it is of utmost importance that we are able to explain how these decisions came about to be able to actually trust them. The goal of this project is to take a peek inside the opaque operations of these machine learning models from neural networks to LLMs and try to understand which features played a key role in the decision-making.
Page updated
Report abuse