RoboCasa365
PulseAugur coverage of RoboCasa365 — every cluster mentioning RoboCasa365 across labs, papers, and developer communities, ranked by signal.
7 day(s) with sentiment data
-
LLM pipeline SimAudit finds 61 defects in AI manipulation benchmark success checkers
Researchers have developed SimAudit, an LLM-based pipeline designed to audit the success checkers used in simulated manipulation benchmarks. These checkers, which label episodes as success or failure, can incorrectly id…
-
New APS-VLA framework accelerates robotic control by pruning visual tokens
Researchers have developed APS-VLA, a novel framework designed to enhance the efficiency of vision-language-action (VLA) models used in robotic manipulation. This method focuses on pruning visual tokens, which are often…
-
New robot learning systems EmbodiedRSI and SPIRE enhance continual learning capabilities
Researchers have developed two novel approaches to enhance embodied continual learning in robots. EmbodiedRSI utilizes a dual-system architecture with a hypothesis graph to guide physical experiments, leading to improve…
-
Auriga agent harness advances zero-shot robot manipulation
Researchers have developed Auriga, a new task-agnostic agent harness designed to improve zero-shot robot manipulation by bridging the gap between planning and execution. Auriga integrates explicit task beliefs, precise …
-
OmniBrain-0: Unified Embodied Agent for Robots Unveiled
Researchers have introduced OmniBrain-0, a new unified embodied agent designed to coordinate manipulation and navigation tasks for robots. This lightweight model integrates task understanding, planning, progress trackin…
-
Patchwork system slashes simulation resource use by 92%
Researchers have developed Patchwork, a novel system designed to optimize resource sharing in domain-randomized embodied simulations. This system separates common scene resources into shared templates and specific mutat…
-
RynnUnify: A Unified Embodied Model for Planning and Robot Control
Researchers have introduced RynnUnify, a 16-billion parameter unified embodied model designed to integrate visual planning, physical reasoning, and robot control. This Mixture-of-Transformers architecture builds upon Co…
-
New framework CausalAlign enhances video generation model ego-motion consistency
Researchers have developed CausalAlign, a new framework designed to improve the consistency of camera ego-motion in diffusion-based video generation models (VGMs). The framework uses causal interventions to identify how…
-
New CausalAlign framework boosts video generation model ego-motion consistency
Researchers have developed CausalAlign, a new framework designed to improve the consistency of camera ego-motion in diffusion-based video generation models. The framework addresses the issue of ego-motion drift over lon…
-
New benchmark evaluates VLA policy robustness in household tasks
Researchers have introduced RoboCasa-Shift, a new benchmark designed to evaluate the robustness of vision-language-action (VLA) policies in complex household manipulation tasks. This benchmark moves beyond simple aggreg…
-
LoopWAM model improves robot policies using mixed-quality deployment data
Researchers have developed LoopWAM, a closed-loop world-action model designed to improve generalist robot policies by learning from mixed-quality deployment rollouts. Unlike methods that filter out failures, LoopWAM uti…
-
New REKI VLA Model Achieves Strong Performance-Size Balance
Researchers have introduced REKI, a 2.4 billion parameter vision-language-action model (VLA) that achieves a strong balance between performance and size. Trained solely on public data, REKI demonstrates competitive resu…
-
New robotic world action models improve efficiency and performance · 6 sources tracked
Researchers have developed new methods for world action models (WAMs) in robotics. RAP-WAM uses predictive rehearsal and action perturbations to improve policy learning, achieving high success rates on real-world tasks.…
-
EmbodiedRSI system enhances robot foundation models through autonomous skill evolution
Researchers have developed EmbodiedRSI, a novel self-evolving agentic harness designed to improve robot foundation models. This system autonomously selects experiments to refine code and skills, addressing the challenge…
-
New framework cuts robot agent token use, boosts success rate
Researchers have developed PyRUA-Lean, a new framework designed to optimize the efficiency of robot agents controlled by vision-language models (VLMs). This framework reduces token overhead by composing robot primitives…
-
New research advances robot world models for efficient planning and control · 10 sources tracked
Recent research explores advancements in world action models (WAMs) for robotics and agent planning. WAM-Cache and WAMJET focus on accelerating inference by optimizing key-value reuse and employing agentic harnesses for…
-
GPT-6 Astra evaluated as embodied robot policy across six domains
A new paper explores the capabilities of GPT-6 Astra, a model designed for embodied policies in robotics. The research evaluates Astra's performance across six domains, including manipulation, mobile manipulation, navig…
-
New frameworks enhance mobile manipulation with improved data efficiency and robustness · 4 sources tracked
Researchers are developing advanced frameworks for mobile manipulation in robotics, aiming to improve data efficiency and robustness. VOMMI utilizes portable demonstrations and offline trajectory reconstruction to enhan…
-
New VLA models enhance robot reasoning, safety, and efficiency · 10 sources tracked
Researchers are developing advanced Vision-Language-Action (VLA) models for robotics, focusing on improving reasoning, safety, and efficiency. New methods like TOAST and TRUST aim to enhance policy learning and runtime …
-
New VLA models enhance robot manipulation with foresight and large-scale data
Researchers have developed new methods to improve the performance of Vision-Language-Action (VLA) models in robotics, particularly for complex, long-horizon tasks. One approach, Foresight Residual RL, enhances credit as…