Research
-
TL;DR: GEMA turns a robot's past observations into task-specific guidance: a multimodal language model, acting as a memory foundation model, interprets what the robot saw to generate text-and-image references (or reuses a retrieved observation directly), which a world action model executes while a task agent picks subtasks and checks progress. On three memory-dependent manipulation tasks it averages 81.0% success, 12 points above the strongest memory-augmented baseline.
-
TL;DR: Generalized planning interfaces let agents specify what a robot should achieve and how it should act, at any level of abstraction. SPROUT supplies the physical variables and constraints needed for execution, raising success by 29 percentage points over less flexible agent-robot interfaces.
-
TL;DR: KeyFlow adapts a robot policy without retraining: a coding agent picks a few semantic 3D keypoints, writes short programs that turn an episodic memory of ten demonstrations into a flow field the robot follows, and revises both from its own play rollouts. On 13 RoboCasa tasks it beats π0.5 fine-tuned on ten times as many demonstrations (62.6% vs. 47.8%), and in IsaacSim its flow fields raise a base π0.5 policy's success by 37.2 points.
-
TL;DR: This paper introduces a framework for few-shot learning from demonstration by inferring the embodied reasoning process behind observed actions. On challenging reasoning tasks, our method achieves strong generalization to new tasks with only 1-3 demonstrations, moving closer towards human performance.
-
TL;DR: This work introduces a hierarchical and modular vision-language architecture for multi-step humanoid manipulation tasks. Specifically, we leverage vision-language models (VLMs) for high-level planning and action-chunking transformer (ACT) and whole-body RL for mid and low-level control.
-
TL;DR: This paper explores the use of vision foundational models and pre-trained representations to enhance tactile representation learning and multi-modal continual learning.
-
TL;DR: This work presents a method for simultaneously pruning and quantizing neural networks by approximately minimizing the distance in parameter space.
-
TL;DR: This paper introduces a Fisher Information Metric approximation method for model sensitivity to low-bit quantization.