Agility A2 ExoArm 7: A Powerful Platform for Bimanual Teleoperation and Robot Learning

Introduction

As robotics research rapidly advances toward embodied intelligence and autonomous manipulation, the demand for flexible and high-performance robotic platforms continues to grow. Researchers and developers require systems capable of collecting high-quality data, performing complex manipulation tasks, and supporting advanced AI algorithms.

The Agility A2 ExoArm 7 Bimanual Teleoperation Kit is designed to meet these needs. Built as a high-payload dual-arm robotic platform, it provides researchers with a powerful tool for teleoperation, robot learning, and embodied AI development. With its open architecture, robust hardware design, and advanced software ecosystem, Agility A2 bridges the gap between laboratory experimentation and real-world robotic applications.

High-Performance Dual-Arm Architecture

The Agility A2 features a bimanual robotic configuration, with each arm equipped with 7 degrees of freedom (DOF). This design allows the robot to replicate human-like arm movements and perform complex manipulation tasks requiring coordination between two arms.

Each arm offers a 714 mm reach, enabling a wide workspace suitable for various laboratory and industrial scenarios. The platform is also capable of handling demanding workloads, with a rated payload of 5 kg per arm and a peak payload of up to 12 kg, making it powerful enough to manipulate real tools and industrial components.

Constructed using aluminum alloy and stainless steel, the robot is designed for durability and long-term operation. Combined with Quasi-Direct Drive (QDD) compliant joints and a 48 V drive system, the Agility A2 delivers smooth, responsive, and human-like motion, which is essential for precise manipulation and teleoperation tasks.

Built for Teleoperation and Data Collection

One of the key strengths of the Agility A2 is its ability to support advanced teleoperation systems. The platform can connect with external leader devices such as VR controllers, exoskeleton interfaces, or master robotic arms, forming a leader-follower control system for intuitive remote operation.

Through teleoperation, researchers can collect high-quality demonstration data for machine learning models. This data is particularly valuable for developing algorithms based on:

• Imitation learning

• Reinforcement learning

• Vision-Language-Action (VLA) models

By enabling human operators to directly control the robot during tasks such as grasping, sorting, and assembly, the Agility A2 provides an efficient method for generating training datasets for embodied AI systems.

Open Software Ecosystem for Robotics Development

The Agility A2 is designed with research flexibility in mind. It ships with a complete software environment that includes Ubuntu 22.04 and ROS2 Humble, allowing seamless integration with modern robotics frameworks.

The system also supports NVIDIA Isaac, making it easier to bridge simulation and real-world deployment for AI models. This compatibility enables researchers to train models in simulation and then deploy them directly onto the physical robot.

Additionally, the platform includes the OpenArmX SDK, which provides essential tools for motor management, diagnostics, simulation, and teleoperation. These tools significantly reduce development time and allow robotics teams to focus on algorithm development rather than low-level system integration.

Versatile Applications Across Research and Industry

Thanks to its open architecture and powerful manipulation capabilities, the Agility A2 can be used in a wide range of research and development scenarios.

Embodied AI and Robot Learning

For universities and research institutes, the Agility A2 serves as an ideal platform for studying robot learning and embodied intelligence. Researchers can evaluate machine learning models on real-world manipulation tasks such as object sorting, grasping, and dual-arm coordination.

Remote Manipulation in Hazardous Environments

In laboratories simulating nuclear, chemical, or biosafety workflows, the robot can function as a remote manipulator. Operators can control the arms to turn valves, press switches, and transfer hazardous materials while maintaining a safe distance.

Industrial Robotics Research

For industrial R&D teams, the Agility A2 enables rapid prototyping of dual-arm robotic workcells. The system can perform contact-rich tasks such as tightening screws, plugging connectors, or assembling components, making it valuable for testing human-robot collaboration strategies before deploying industrial automation systems.

Robotics Education and Training

The platform is also well suited for robotics education and hands-on training. Its open architecture allows students to explore topics such as robot kinematics, ROS2 control, teleoperation, and AI-driven manipulation using professional-grade hardware.

Flexible Configuration Options

The Agility A2 system is available in multiple configurations to suit different research needs.

A standard package includes the dual-arm robotic system, an industrial control computer, and a complete cable harness. For advanced teleoperation experiments, an isomorphic leader-follower configuration is also available, allowing one robotic arm to act as the control interface while the other performs the task.

These configuration options make it easy for laboratories and research teams to build a complete teleoperation and robot learning pipeline.

Conclusion

The Agility A2 ExoArm 7 Bimanual Teleoperation Kit represents a powerful and flexible platform for modern robotics research. By combining high-payload hardware, an open software ecosystem, and advanced teleoperation capabilities, it enables researchers to explore new frontiers in embodied AI, robot learning, and human-robot collaboration.

Whether used for academic research, industrial prototyping, or robotics education, the Agility A2 provides a scalable and reliable solution for building the next generation of intelligent robotic systems.