Axol

Product Introduction

1. Definition: Axol is a dual-arm collaborative robot (cobot) platform specifically engineered for physical AI development and deployment. It is a fully integrated, 7-degree-of-freedom (7-DOF) per arm robotic system designed for bimanual manipulation tasks, featuring high-resolution sensors and an open-source software stack.
2. Core Value Proposition: Axol exists to dramatically lower the barrier for automating real-world, complex manipulation tasks using physical AI. It provides a powerful, extensible, and "ready-out-of-the-box" foundation for teams building, training, and deploying AI-powered robot policies, eliminating the need to build a custom hardware platform from scratch.

Main Features

1. Extra-Long 860mm Arm Reach: Each of Axol's arms provides an industry-leading 860mm of reach from shoulder to fingertip. This is achieved through optimized kinematic design and 7 independent joints per arm. This extended reach grants a significantly larger continuous workspace, reducing the need for the robot to be repositioned or for tasks to be split between multiple setup positions, thereby increasing operational efficiency and enabling larger-scale tasks.
2. Reduced Singularities & 7-DOF Mobility: Axol features full 180° pitch and yaw rotation at the wrist joint. This advanced kinematic design, combined with 7 degrees of freedom per arm, dramatically minimizes the number of shoulder singularities—common kinematic deadspots in robotic arms that cause loss of mobility. The result is a larger, more effective operational workspace and smoother, more reliable trajectories for complex paths, which is critical for data collection and policy execution in physical AI.
3. Open-Source SDK & Bimanual Control Stack: Axol ships with a comprehensive Python SDK and CLI that includes a built-in bimanual inverse kinematics (IK) solver, low-level CAN motor interface, and bindings for popular frameworks like LeRobot. This open-source stack provides everything from raw joint control to policy training pipelines. It is designed to be extended, allowing researchers and developers to integrate their own perception, planning, and learning algorithms directly onto the platform.
4. WebXR VR Teleoperation & Data Collection: The system includes a WebXR-based virtual reality teleoperation interface. From any compatible headset, operators can stream hand and elbow poses over WebSocket to control Axol directly. This mode features built-in tools for high-quality, labeled data collection and recording, which is the essential first step for training imitation learning and other physical AI models with real-world data.
5. Deployment-Ready & Modular Construction: Built with steel, aluminum, and TPU, Axol is constructed for long-term reliability in production environments. Its modular design allows for easy integration with third-party hardware, such as mobile bases or custom lifts. Features like FAKRA GMSL 2.0 passthrough connectors on the wrists enable high-speed, low-latency connections for external machine vision cameras (like the ZED X), essential for visual servoing and AI perception.

Problems Solved

1. Pain Point: The high cost, complexity, and lead time associated with developing or acquiring a custom dual-arm robotic platform for AI research and automation prototyping. Teams spend significant time and capital on mechanical integration, software driver development, and low-level control system design before any AI work can begin.
2. Target Audience: This product is built for AI/Robotics Researchers, Automation Engineers, Robotics Software Developers, and R&D Teams in Manufacturing/Logistics who need a capable, programmable, dual-arm hardware substrate to develop and validate physical AI policies without embarking on a multi-month hardware integration project.
3. Use Cases: Essential for developing and training contact-rich manipulation policies (e.g., assembly, packaging, lab automation), creating high-quality robotic manipulation datasets for imitation learning, rapid proof-of-concept prototyping for new bimanual automation tasks, and serving as the core hardware platform for physical AI research in academic and commercial labs.

Unique Advantages

1. Differentiation vs. Comparable Platforms: Compared to other arms in its price bracket (like the YAM or WidowX), Axol offers a superior combination of longer reach (860mm), higher peak payload (6.5kg), and more kinematic freedom (7-DOF with reduced singularities). While many robots offer single arms or more limited motion, Axol provides a complete, integrated bimanual system with a unified control stack, outperforming on core specifications that directly impact real-world task capability.
2. Key Innovation: The key innovation is the complete fusion of high-performance, deployment-ready dual-arm hardware with a mature, open-source physical AI software ecosystem. Unlike generic cobots requiring extensive custom programming, or research platforms lacking robust construction, Axol is purpose-built for the AI-first development loop (data collection -> training -> deployment) and ships with the necessary SDK, teleoperation tools, and hardware interfaces to support it immediately.

Frequently Asked Questions (FAQ)

1. What is the price and what is included in the Axol Kit? The Axol Kit, which includes everything needed to start building and training AI-powered manipulation applications, is priced at $11,999 (launch discount). It contains the Axol dual-arm robot, a height-adjustable Axol Base, three ZED X One S cameras, a ZED Box Orin NX 16GB, and all necessary FAKRA GMSL 2.0 cables. The standalone Axol robot is available for $7,999.
2. Is Axol compatible with ROS (Robot Operating System)? While Axol does not explicitly state ROS compatibility in the provided material, its open-source Python SDK and CLI with bimanual IK and low-level CAN interface provides a solid foundation for integration. Developers can use this SDK to build their own ROS nodes or middleware bridges to incorporate Axol into a ROS-based workflow if needed.
3. What are the primary applications for a dual-arm robot like Axol? Axol is designed for physical AI research and real-world automation of complex, contact-rich tasks. Key use cases include bimanual manipulation for assembly and packing, data collection for training robot policies, research into dexterous manipulation, and developing collaborative robots for human-robot interaction scenarios in laboratory or light industrial settings.
4. How does Axol's open-source software support physical AI development? The open-source SDK includes critical tools for the entire AI development pipeline: VR teleoperation for data collection, a bimanual IK solver for motion planning, and LeRobot bindings for connecting to popular learning frameworks. This allows researchers to directly collect real-world demonstration data, train policies on the collected data, and deploy the resulting models back onto the same hardware platform, all within an open and extensible ecosystem.