Scalabot on September 15 launched HERON-Context Reinforcement Action Model (CRA), the second model built on its HERON technical architecture, which features memory, reinforcement learning for error recovery, dexterous and deformable object manipulation, and cross-embodiment generalization to equip robots with continuously evolving real-world operational capabilities, shifting them from task execution to learning and iterative improvement.
Scalabot Unveils HERON-CRA, Its Second Model Under HERON Tech Architecture
Related Articles
Hesai and JD Group Partner to Expand Physical AI Real-world Applications
D-Robotics, GigaAI partner to develop edge embodied AI
Kinetix AI secures over 500 mln yuan angel+ round backed by Temasek-linked Vertex for embodied AI
Musk says SpaceX confident of launching Nvidia VR NLV72 AI computer to orbit next year
Xieyue Intelligence Secures Angel+ Funding to Build Home Embodied AI Foundation Model
2026 Inclusion Bund Summit: Robot Grasps Diverse Objects Without Calibration
China accounts for over 97% of global humanoid robot shipments in H1 2026, data shows
Westlake Robotics secures A1 round financing following August Series A fundraising
the 2026 Bund Summit




