Hiwonder NexArm Open-Source VLA Embodied AI Robotic Arm with ESP32 & AT32 Dual-Chip Design, AI Vision, Large AI Model, and K230 Vision Module

The Hiwonder NexArm is an advanced open-source robotic arm platform designed for students, educators, and researchers exploring embodied AI and Vision-Language-Action (VLA) learning. Powered by a dual-chip design and the LeRobot ecosystem, this versatile arm supports everything from imitation learning to full-scenario deployment.

Embodied AI development platform

The NexArm covers the entire development workflow from data collection and imitation learning training to real-world deployment. It integrates seamlessly with mobile chassis, sliding rails, and conveyor belts to build comprehensive, full-scenario embodied AI systems.

Magnetic encoder bus servos

Built with a dual-output-shaft magnetic encoder servo system and all-metal gears, this arm delivers precise and stable joint movement. A powerful magnetic motor combines with real-time feedback algorithms to ensure highly accurate motion control.

Adaptive inverse kinematics

The robotic arm automatically detects target objects within its recognition area, calculating their exact coordinates and placement angles. It then adaptively adjusts each joint using advanced inverse kinematics algorithms to execute flexible, precise grasping.

Imitation learning and teleoperation

Designed for advanced robotics research, the platform supports a leader-follower teleoperation system using a master and follower arm setup. It features deep, low-level compatibility with the latest Hugging Face LeRobot framework for seamless machine learning integration.

K230 vision processing power

Equipped with a K230 module delivering 6 TOPS of computing power, the arm supports over 30 AI vision functions including YOLO. This allows for multimodal large-model deployment and advanced hand-eye coordination in complex operating environments.