Chelsea Finn

MOTO enhances robot learning by combining offline training with online fine-tuning.

AutoFT improves model performance on unseen data through innovative fine-tuning techniques.

A new system enhances how robots gather and learn from real-world data.

Introducing SERL: A new framework for efficient robot learning tasks.

This article discusses a method to improve LLMs using verbal feedback without overgeneralization.

A study on the effectiveness of RLAIF versus supervised fine-tuning for language models.

A method to enhance machine learning using unlabeled data across various domains.

Study reveals benefits of training robots with varied task experiences.

Robots are learning to adapt and perform tasks across various fields.

Robots learn to adapt and improve by receiving real-time human feedback.

New model improves machine learning by separating complex data into simpler parts.

This study evaluates methods to enhance large language models using user preference data.

This article discusses the role of simulations in robot skill evaluation.

Generalist robot policies enhance adaptability for diverse tasks in robotics.

Examining over-optimization in DAAs and its impact on model performance.

A new model enables robots to learn tasks better using visual and language inputs.

A new approach helps robots adapt quickly to unfamiliar tasks.

New methods improve legged robots' movement in complex environments using AI.

Robots improve navigation by understanding both speech and images.

A new method to assess diverse user values in language models.

This method enhances the reliability of language model confidence scores.

A fun way for everyone to help teach robots through crowdsourced data.

Discover how Policy Agnostic Reinforcement Learning changes machine decision-making.