MoDem: Accelerating Visual Model-Based Reinforcement Learning with Demonstrations

1 code implementation • 12 Dec 2022 • Nicklas Hansen, Yixin Lin, Hao Su, Xiaolong Wang, Vikash Kumar, Aravind Rajeswaran

We identify key ingredients for leveraging demonstrations in model learning -- policy pretraining, targeted exploration, and oversampling of demonstration data -- which forms the three phases of our model-based RL framework.

Model-based Reinforcement Learning reinforcement-learning +1

Transformers are Adaptable Task Planners

no code implementations • 6 Jul 2022 • Vidhi Jain, Yixin Lin, Eric Undersander, Yonatan Bisk, Akshara Rai

Every home is different, and every person likes things done in their particular way.

Differentiable and Learnable Robot Models

1 code implementation • 22 Feb 2022 • Franziska Meier, Austin Wang, Giovanni Sutanto, Yixin Lin, Paarth Shah

Building differentiable simulations of physical processes has recently received an increasing amount of attention.

Translating Robot Skills: Learning Unsupervised Skill Correspondences Across Robots

no code implementations • 29 Sep 2021 • Tanmay Shankar, Yixin Lin, Aravind Rajeswaran, Vikash Kumar, Stuart Anderson, Jean Oh

In this paper, we explore how we can endow robots with the ability to learn correspondences between their own skills, and those of morphologically different robots in different domains, in an entirely unsupervised manner.

Translation Unsupervised Machine Translation

Efficient and Interpretable Robot Manipulation with Graph Neural Networks

no code implementations • 25 Feb 2021 • Yixin Lin, Austin S. Wang, Eric Undersander, Akshara Rai

Manipulation tasks, like loading a dishwasher, can be seen as a sequence of spatial constraints and relationships between different objects.

Imitation Learning Robot Manipulation

Learning State-Dependent Losses for Inverse Dynamics Learning

1 code implementation • 10 Mar 2020 • Kristen Morse, Neha Das, Yixin Lin, Austin S. Wang, Akshara Rai, Franziska Meier

In both settings, the structured and state-dependent learned losses improve online adaptation speed, when compared to standard, state-independent loss functions.

Meta-Learning

Encoding Physical Constraints in Differentiable Newton-Euler Algorithm

1 code implementation • L4DC 2020 • Giovanni Sutanto, Austin S. Wang, Yixin Lin, Mustafa Mukadam, Gaurav S. Sukhatme, Akshara Rai, Franziska Meier

The recursive Newton-Euler Algorithm (RNEA) is a popular technique for computing the dynamics of robots.

Robotics

Curious iLQR: Resolving Uncertainty in Model-based RL

no code implementations • 15 Apr 2019 • Sarah Bechtle, Yixin Lin, Akshara Rai, Ludovic Righetti, Franziska Meier

In this work, we propose a model-based reinforcement learning (MBRL) framework that combines Bayesian modeling of the system dynamics with curious iLQR, an iterative LQR approach that considers model uncertainty.

Model-based Reinforcement Learning reinforcement-learning +1