Awesome Robotics Company Research

Basic Info. This repo collects notes and links about robotics companies. It is organized as a lightweight awesome list, inspired by awesome-humanoid-robot-learning.

Use short entries first. Add deeper notes only when they are useful.

Company Index

• AgiBot, building and deploying general-purpose embodied AI systems, website.
• BeingBeyond, building general-purpose foundational models for robots from large-scale human data, website.
• Figure, building general-purpose humanoid robots for home and workforce applications, website.
• Physical Intelligence, bringing general-purpose AI into the physical world, website.
• Sharpa, building ultra-high performance robots and core components, website.

---

AgiBot

• 2026.04, Learning while Deploying: Fleet-Scale Reinforcement Learning for Generalist Robot Policies, website
• 2026.01, SOP: Scaling General-Purpose Robots in the Real World, website
• 2026.01, Act2Goal: From World Model to General Goal-Conditioned Policy, website
• 2026.01, Unified Embodied VLM Reasoning with Robotic Action via Autoregressive Discretized Pre-training, website

BeingBeyond

• 2026.04, Unmasking the Illusion of Embodied Reasoning in Vision-Language-Action Models, website
• 2026.04, Being-H0.7: A Latent World-Action Model from Egocentric Videos, website
• 2026.03, OpenT2M: No-frill Motion Generation with Open-source, Large-scale, High-quality Data, website
• 2026.02, Joint-Aligned Latent Action: Towards Scalable VLA Pretraining in the Wild, website
• 2026.02, General Humanoid Whole-Body Control via Pretraining and Fast Adaptation, website
• 2026.02, Rethinking Visual-Language-Action Model Scaling: Alignment, Mixture, and Regularization, website
• 2026.01, Being-H0.5: Scaling Human-Centric Robot Learning for Cross-Embodiment Generalization, website
• 2025.12, DemoFunGrasp: Universal Dexterous Functional Grasping via Demonstration-Editing Reinforcement Learning, website
• 2025.12, Spatial-Aware VLA Pretraining through Visual-Physical Alignment from Human Videos, website
• 2025.12, DiG-Flow: Discrepancy-Guided Flow Matching for Robust VLA Models, website
• 2025.10, DemoHLM: From One Demonstration to Generalizable Humanoid Loco-Manipulation, website
• 2025.08, Being-M0.5: A Real-Time Controllable Vision-Language-Motion Model, website
• 2025.07, Being-H0: Vision-Language-Action Pretraining from Large-Scale Human Videos, website
• 2025.06, Being-VL-0.5: Unified Multimodal Understanding via Byte-Pair Visual Encoding, website
• 2024.10, Being-VL-0: From Pixels to Tokens: Byte-Pair Encoding on Quantized Visual Modalities, website

Figure

• 2026.01, Introducing Helix 02: Full-Body Autonomy, website
• 2025.10, Introducing Figure 03, website
• 2025.09, Project Go-Big: Internet-Scale Humanoid Pretraining and Direct Human-to-Robot Transfer, website
• 2025.07, F.03 Battery Development, website
• 2025.06, Scaling Helix: a New State of the Art in Humanoid Logistics, website
• 2025.03, Natural Humanoid Walk Using Reinforcement Learning, website
• 2025.02, Helix Accelerating Real-World Logistics, website
• 2025.02, Helix: A Vision-Language-Action Model for Generalist Humanoid Control, website

Physical Intelligence

• 2026.04, π 0.7: a Steerable Model with Emergent Capabilities, website
• 2026.03, Precise Manipulation with Efficient Online RL, website
• 2026.03, VLAs with Long and Short-Term Memory, website
• 2026.02, The Physical Intelligence Layer, website
• 2025.12, Moravec's Paradox and the Robot Olympics, website
• 2025.12, Emergence of Human to Robot Transfer in VLAs, website
• 2025.11, π*0.6: a VLA that Learns from Experience, website
• 2025.06, Real-Time Action Chunking with Large Models, website
• 2025.05, VLAs that Train Fast, Run Fast, and Generalize Better, website
• 2025.04, π 0.5: a VLA with Open-World Generalization, website
• 2025.02, Teaching Robots to Listen and Think Harder, website
• 2025.02, Open Sourcing π 0, website
• 2025.01, FAST: Efficient Robot Action Tokenization, website
• 2024.10, π 0: Our First Generalist Policy, website

Sharpa

• 2026.03, Towards Human-Like Manipulation through RL-Augmented Teleoperation and Mixture-of-Dexterous-Experts VLA, website
• 2026.03, DexEMG: Towards Dexterous Teleoperation System via EMG2Pose Generalization, website
• 2026.02, Tacmap: Bridging the Tactile Sim-to-Real Gap via Geometry-Consistent Penetration Depth Map, website
• 2026.02, EgoScale: Scaling Dexterous Manipulation with Diverse Egocentric Human Data, website
• 2026.02, SimToolReal: An Object-Centric Policy for Zero-Shot Dexterous Tool Manipulation, website
• 2026.01, Spatially anchored Tactile Awareness for Robust Dexterous Manipulation, website