RoboMimic
PulseAugur coverage of RoboMimic — every cluster mentioning RoboMimic across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
MAC-v2 framework enhances pixel-based offline reinforcement learning
Researchers have introduced MAC-v2, a novel framework designed for long-horizon model-based value expansion in pixel-based offline reinforcement learning. This approach tackles the inherent difficulty of learning both r…
-
New framework CuraEval optimizes robot imitation learning data selection
Researchers have developed CuraEval, a framework designed to improve imitation learning in robotics by optimizing the selection of training data. The framework analyzes demonstration scores to identify the most influent…
-
New robot policy allows real-time control over speed and smoothness
Researchers have developed a Quality-Controllable Policy (QCP) for robots, enabling dynamic control over execution attributes like speed and smoothness at inference time. To expand the range of controllable qualities be…
-
New robot policy allows real-time quality control and iterative improvement
Researchers have developed a Quality-Controllable Policy (QCP) that allows robots to adjust their execution quality in real-time, such as speed or smoothness. This policy improves performance through an iterative hindsi…
-
Robotics research introduces feedback world model for enhanced policy precision
Researchers have developed a "feedback world model" that enhances the precision of diffusion policies in robotics. This novel inference-time mechanism iteratively corrects predictions using real-time observations, compe…
-
New QF3 algorithm trains robot locomotion policies 10x faster
Researchers have introduced QF3 (Fast Flow RL with Filtered Q-Gradients), a novel off-policy reinforcement learning algorithm designed to train robot policies more efficiently. QF3 utilizes flow matching combined with t…
-
New imitation learning methods enhance robot reliability and performance · 4 sources tracked
Researchers are developing new methods to improve imitation learning for robots, focusing on enhancing reliability and performance with imperfect data. Rewind-IL introduces a framework for online failure detection and r…
-
New QWM Framework Enhances Reinforcement Learning with World Models
Researchers have introduced QWM, a novel framework that integrates world models with Q-learning to enhance sample efficiency in reinforcement learning. This approach uses world models for test-time search over imagined …
-
DriftWorld model accelerates robotic planning with faster world modeling · 2 sources tracked
Researchers have developed DriftWorld, a novel action-conditioned world model that significantly accelerates robotic planning. Unlike diffusion-based models that require iterative denoising, DriftWorld uses a single for…
-
ReGuide framework improves diffusion policies with self-improving guidance
Researchers have developed ReGuide, a novel framework designed to improve the performance of behavior-cloned diffusion policies, which are often susceptible to covariate shift. ReGuide utilizes Phase-Conditioned Guidanc…
-
RE4 framework enhances object interaction imitation learning
Researchers have introduced RE4, a new framework for imitation learning in object interaction tasks. RE4 aims to balance performance and interpretability by repurposing established manipulation theories. The framework i…
-
VLMs improved for world modeling via inverse dynamics prediction
Researchers are exploring methods to improve the predictive capabilities of vision-language models (VLMs) for world modeling. A key challenge is that VLMs struggle with forward dynamics prediction (generating future sta…
-
New research advances policy optimization for robotics and LLMs
Researchers have introduced several new methods to enhance policy optimization in reinforcement learning, particularly for complex tasks involving robotics and large language models. MODIP aims to efficiently fine-tune …
-
New GAP method boosts robotic manipulation learning with scarce data
Researchers have developed Geometric Anchor Pre-training (GAP), a novel method to improve data efficiency in visuomotor learning for robotic manipulation. GAP pre-trains a spatial adapter to generate stable geometric an…
-
New flow map policies accelerate generative AI for robotics
Researchers have developed a new class of generative policies called flow map policies, designed to accelerate action generation in complex control problems. These policies learn to make large jumps within generative dy…
-
StereoPolicy enhances robot manipulation with stereo vision
Researchers have developed StereoPolicy, a new framework designed to enhance robotic manipulation by utilizing synchronized stereo image pairs. This approach strengthens geometric reasoning for robots, overcoming the de…
-
STEP method accelerates diffusion policies for faster, more accurate robot control
Researchers have introduced STEP, a novel method to accelerate diffusion policies for visuomotor control in robotics. STEP utilizes spatiotemporal consistency prediction to generate high-quality warm-start actions, sign…