RoboMME
PulseAugur coverage of RoboMME — every cluster mentioning RoboMME across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
New robotic agent SERA improves long-horizon manipulation with self-evolving plans
Researchers have developed SERA, a Self-Evolving Robotic Agent designed to improve long-horizon robotic manipulation tasks. SERA utilizes an editable harness to maintain task context and evolve its planning strategies b…
-
New robot memory system encodes episodes as fixed-size neural representations
Researchers have developed a new memory framework for robots called inr2action, which encodes entire episodes as fixed-size neural implicit representations. This approach allows robots to store and recall information re…
-
New benchmarks and memory methods boost VLA models for long-horizon robotics
Researchers are developing new benchmarks and methods to improve the memory capabilities of vision-language-action (VLA) models for long-horizon robotic manipulation tasks. These new approaches aim to address the challe…
-
PonderPounce uses MLLM as robot memory for enhanced control
Researchers have developed PonderPounce, a novel system that leverages a pretrained multimodal large language model (MLLM) as an episode context engine for robot control. Instead of relying on purpose-built memory modul…
-
PonderPounce: MLLMs leverage native context for robot control memory
Researchers have introduced PonderPounce, a novel approach that utilizes the inherent causal context of multimodal large language models (MLLMs) as a memory system for robot control. This method jointly trains a reasoni…
-
New benchmark reveals robot memory struggles with interference
Researchers have introduced RoboMME-Interference, a new benchmark designed to evaluate the long-context memory capabilities of robots, particularly their performance under distracting conditions. The benchmark, built up…
-
New benchmark reveals current robot memory systems fail under interference
Researchers have introduced RoboMME-Interference, a new benchmark designed to test robot memory capabilities in realistic, long-term scenarios with distracting information. The benchmark reveals that current robot memor…
-
New benchmark reveals robot memory struggles with long-term recall and interference
Researchers have introduced RoboMME-Interference, a new benchmark designed to test robot memory systems under conditions that mimic real-world deployment, where robots accumulate experience over extended periods and acr…
-
New AI models enhance robot manipulation with advanced memory systems · 4 sources tracked
Researchers have introduced two new methods for improving robot manipulation through enhanced memory systems. Mem-World, a memory-augmented multi-view action-conditioned world model, addresses challenges in persistent w…
-
New DIRECT framework optimizes compute for embodied AI planners
Researchers have developed a new framework called DIRECT to optimize the allocation of computational resources for embodied AI planners. The system analyzes multimodal scene context to intelligently route compute, impro…