paper

Geometric Action Model for Robot Policy Learning

paperactiveprovisionalgeometric-action-model-for-robot-policy-learning-5a9205b3·1 events·first seen 36h ago

Aliases: Geometric Action Model for Robot Policy Learning

Co-occurring entities

More like this (12)

Geometric Action Model Learning Red Agent Policy from Observations for Neurosymbolic Autonomous Cyber Agents TempoVLA: Learning Speed-Controllable Vision-Language-Action Policies APPO: Agentic Procedural Policy Optimization Preference Coordinated Multi-agent Policy Optimization Reward Modeling for Multi-Agent Orchestration Vector Policy Optimization Visual Verification Enables Inference-time Steering and Autonomous Policy Improvement Difference-Aware Retrieval Policies for Imitation Learning An Agency-Transferring Model-Free Policy Enhancement Technique UniIntervene: Agentic Intervention for Efficient Real-World Reinforcement Learning Evolved Policy Gradients

Recent events (1)

6arXiv · cs.LG·36h ago·source ↗

Geometric Action Model (GAM) repurposes geometric foundation models for 3D-aware robot manipulation

Researchers propose the Geometric Action Model (GAM), a language-conditioned robot manipulation policy that splits a pretrained geometric foundation model (GFM) to serve simultaneously as an observation encoder, causal future predictor, and action decoder. Unlike existing vision-language-action models that operate on 2D image frames, GAM explicitly incorporates 3D geometric priors for contact-rich manipulation. The approach claims improvements in accuracy, robustness, speed, and model size over foundation-model-scale baselines across simulation and real-robot benchmarks.

Agent and Tool Ecosystem Multimodal Progress Geometric Action Model for Robot Policy Learning Geometric Action Model