Free Reads
Sign in to view your remaining parses.
Tag Filter
Large-Scale Robot Demonstration Dataset
OpenVLA: An Open-Source Vision-Language-Action Model
Published:6/13/2024
Open-Source Vision-Language-Action ModelRobotic Multi-Task ManipulationLarge-Scale Robot Demonstration DatasetModel Fine-Tuning and AdaptationVision-Language-Action Model
OpenVLA is a 7Bparameter opensource VisionLanguageAction model built on Llama 2. Training on 970k robot demonstrations, it efficiently finetunes for various tasks, outperforming RT2X by 16.5% in absolute success rate across 29 tasks.
03
Octo: An Open-Source Generalist Robot Policy
Published:5/21/2024
Generalist Robot PoliciesMulti-modal action representation and modelingTransformer architectureLarge-Scale Robot Demonstration DatasetRobotic Action Learning
Octo is an opensource transformerbased generalist robot policy pretrained on 800K trajectories, enabling fast finetuning across diverse sensors and robots, guided by language or images, demonstrating strong generalization on nine platforms.
06
UMI-on-Air: Embodiment-Aware Guidance for Embodiment-Agnostic Visuomotor
Policies
LLM-guided motion planningMultimodal Robot LearningMulti-modal action representation and modelingLarge-Scale Robot Demonstration DatasetGeneralist Robot Policies
UMIonAir uses human demonstrations and EmbodimentAware Diffusion Policy (EADP) to guide visuomotor policies for constrained robot forms, enhancing adaptability, success, and robustness across different embodiments.
07