New Product Launch: Agility A2 Flex – A 27-DOF Wheeled Dual-Arm Robot for Embodied AI and VR Teleoperation

As embodied AI and robotic automation continue to evolve, researchers and developers face a growing challenge: how can robots move freely, manipulate objects, and coordinate multiple actions within a real-world environment?

Traditional robotic arms are often limited to fixed workstations, while mobile robots may lack the manipulation capabilities needed for more complex tasks. Combining mobility, dual-arm coordination, perception, and intelligent control into one system can require significant development effort.

Introducing the new Agility A2 Flex, a 27-DOF open-source wheeled dual-arm robot designed to bring mobility, manipulation, and whole-body VR teleoperation together in one integrated platform.

Available in Agility A2 Flex and Agility A2 Flex Ultra configurations, the system provides a flexible foundation for embodied AI research, robotic manipulation, human demonstration data collection, and robot learning.

What Problem Does Agility A2 Flex Solve?

For robotics developers, building a robot that can move and perform manipulation tasks often involves integrating several independent systems, including robotic arms, mobile platforms, sensors, motion controllers, and software frameworks.

This process can be time-consuming, particularly when coordinated movement between different components is required.

Agility A2 Flex addresses this challenge with an integrated 27-DOF whole-body robotic architecture. It combines a mobile base, vertical lifting mechanism, dual robotic arms, grippers, and a movable head within one coordinated control system.

Instead of developing each component separately, researchers can work with a unified robotic platform that supports whole-body movement and manipulation.

This provides a practical starting point for developing, testing, and improving robotic behaviors in dynamic environments.

How Does Agility A2 Flex Handle Tasks at Different Locations and Heights?

A fixed robotic arm can perform repetitive operations within a limited workspace, but many real-world tasks require robots to move between locations and interact with objects at different heights.

Agility A2 Flex combines a 4W4S (Four-Wheel Drive, Four-Wheel Steering) mobile base with a 700 mm vertical lifting mechanism, allowing the robot to adjust both its position and working height.

The multidirectional mobile base enables flexible movement in different directions, while the lifting system helps the robot reach objects positioned at varying heights.

Its two 7-joint robotic arms, each offering an approximate 714 mm operating range, support coordinated bimanual manipulation.

This combination is useful for researching tasks such as object picking, item transfer, coordinated handling, and manipulation across different work surfaces.

By integrating mobility and manipulation, Agility A2 Flex allows developers to explore robotic workflows that go beyond a single fixed workstation.

How Does VR Teleoperation Make Robot Control More Intuitive?

Controlling a multi-degree-of-freedom robot through conventional interfaces can be complicated, especially when the operator needs to coordinate the arms, grippers, head, mobile base, and lifting mechanism.

Agility A2 Flex simplifies this process through whole-body VR teleoperation.

The system includes a Meta Quest 3S VR headset, allowing operators to control different parts of the robot through an integrated teleoperation interface.

Rather than controlling each component through separate systems, users can coordinate whole-body actions in a more intuitive manner.

This capability is particularly valuable for robot demonstration, remote manipulation research, and human-in-the-loop development.

It also provides a practical way to collect human demonstration data, allowing researchers to study how operator-guided movements can contribute to future robotic learning workflows.

How Does Agility A2 Flex Support Embodied AI and Robot Learning?

One of the biggest challenges in embodied AI development is connecting robotic hardware with perception, motion planning, and learning algorithms.

Researchers need a platform that supports repeatable experiments, human demonstration collection, and integration with existing robotics software.

Agility A2 Flex is built on Ubuntu 22.04 and ROS 2 Humble, providing an open development environment for robotics research.

Both configurations support ROS 2 real-time joint control and MoveIt 2 arm motion planning. These tools help developers control robotic joints, plan manipulation movements, and build customized applications.

For teams focusing on advanced embodied AI development, Agility A2 Flex Ultra extends these capabilities with additional perception hardware, navigation support, and a VLA configuration.

The Ultra configuration uses π0.5 as its default VLA model, supporting model training, inference, and workflow validation.

This allows researchers to explore the connection between visual perception, learned action policies, and physical robot execution using an integrated development platform.

Why Is Multi-Sensor Perception Important for Robotic Manipulation?

Robots operating in dynamic environments need to understand both their surroundings and the objects they interact with.

Visual perception helps a robot identify workspace conditions, while spatial sensing provides additional information for positioning and navigation.

Agility A2 Flex includes a head-mounted Intel RealSense D435i camera for visual perception.

For applications requiring more comprehensive environmental sensing, Agility A2 Flex Ultra expands the sensor configuration with dual Intel RealSense D405 wrist cameras, a Livox MID-360S LiDAR, and an additional base-mounted RealSense D435i camera.

The wrist-mounted cameras provide close-range visual information around the manipulation area, while the LiDAR and base-mounted camera provide additional spatial information for navigation and environmental awareness.

Together with Nav2 navigation support, these features make the Ultra configuration more suitable for developing and validating perception-driven mobile manipulation workflows.

How Can Customers Choose Between Agility A2 Flex and Flex Ultra?

Different robotics projects require different levels of computing power, perception, and AI integration.

To accommodate these requirements, Agility A2 Flex is available in two configurations.

Agility A2 Flex – For Teleoperation and Robotics Development

The standard Agility A2 Flex is designed for customers who need a mobile dual-arm robotic platform for VR teleoperation, motion control development, and human demonstration data collection.

It includes an X86 controller, head-mounted RealSense D435i camera, ROS 2 joint control, MoveIt 2 motion planning, and whole-body VR teleoperation.

This configuration is suitable for universities, robotics laboratories, and development teams that want to study mobile manipulation and build their own control applications.

The standard version does not include a VLA configuration.

Agility A2 Flex Ultra – For Advanced Perception and VLA Research

Agility A2 Flex Ultra is designed for teams that require additional sensing, computing resources, navigation, and AI model development capabilities.

It adds an NVIDIA RTX 3090 GPU, wrist-mounted depth cameras, LiDAR, an additional base-mounted camera, and Nav2 navigation.

More importantly, the Ultra version includes a VLA configuration for model training, inference, and workflow validation.

For research teams developing embodied AI policies or exploring perception-driven robotic manipulation, Flex Ultra provides a more comprehensive development environment.

Can Agility A2 Flex Be Expanded for Different Research Needs?

Robotics research often requires customized hardware, end-effectors, sensors, and software environments.

Agility A2 Flex is designed with an open and expandable architecture to support these requirements.

In addition to its standard VR teleoperation system, the platform can be paired with the optional ExoArm-7 Pro wearable exoskeleton to explore an alternative method of human-robot teleoperation.

Optional dexterous hands can also expand manipulation research beyond standard gripper-based tasks. The platform supports custom end-effectors, additional vision devices, and OEM/ODM customization.

These expansion options help developers adapt the robot to different experimental requirements without being limited to a single hardware configuration.

ExoArm-7 Pro and dexterous hands are optional accessories sold separately.

Who Is Agility A2 Flex Designed For?

Agility A2 Flex is primarily designed for robotics researchers, universities, AI developers, and technology companies exploring more advanced robotic systems.

For academic institutions, the platform provides an integrated environment for studying robotics, motion planning, VR teleoperation, and human-robot interaction.

For embodied AI researchers, it supports human demonstration data collection and, with the Ultra configuration, VLA model development and validation.

For robotics companies and R&D teams, its mobile base, dual-arm manipulation system, and expandable hardware architecture provide a foundation for testing new robotic applications and validating proof-of-concept workflows.

Its combination of mobility, manipulation, and open development tools makes it particularly relevant to teams working toward more capable mobile robotic systems.

Conclusion: Bringing Mobile Manipulation and Embodied AI Development Together

Developing robots that can move, perceive, and manipulate objects requires close coordination between mechanical systems, sensors, controllers, and intelligent software.

The new Agility A2 Flex brings these essential capabilities into a unified 27-DOF wheeled dual-arm platform.

With multidirectional mobility, adjustable working height, whole-body VR teleoperation, open-source development support, and expandable hardware options, it helps robotics teams reduce integration complexity and focus on developing practical robotic capabilities.

For customers exploring advanced embodied AI and VLA applications, Agility A2 Flex Ultra further extends the platform with enhanced perception, navigation, GPU computing, and integrated VLA development capabilities.

From human-guided teleoperation to AI-driven robotic research, Agility A2 Flex provides a flexible foundation for the next generation of mobile manipulation development.