Meet AthenaZero: The Two-Armed Robot That Can Throw, Catch, and Hit Baseballs Like a Human

Robots are becoming faster, stronger, and smarter, but matching the natural movement of the human body remains a major challenge. While modern robots can perform repetitive industrial tasks with impressive accuracy, many still struggle to react quickly to moving objects, adjust to unexpected forces, and coordinate their movements in complex situations.

Now, researchers have developed a new two-armed robot called AthenaZero that can throw balls, catch them, and hit them with a baseball bat. Inspired by the movements of human athletes, the robot demonstrates how a different approach to mechanical design could help machines handle objects more naturally and efficiently.

Developed by roboticists at the Robotics and AI Institute (RAI) in Cambridge, Massachusetts, AthenaZero represents a promising step toward robots that can perform fast, coordinated, and physically demanding tasks in real-world environments.

The research was published in the scientific journal Science Robotics in 2026.

Why Is Humanlike Movement So Difficult for Robots?

Human beings can perform a remarkable variety of physical tasks without consciously calculating every movement. We can catch a moving ball, adjust our grip when an object slips, and change the force of our movements depending on what we are handling.

For robots, these actions are much more complicated.

Traditional industrial robots are generally designed to be strong, stiff, and highly precise. These characteristics make them excellent for repetitive jobs such as assembling machinery, moving heavy materials, and handling products on factory production lines.

However, the same stiffness can become a disadvantage when a robot must interact with objects that are moving, fragile, soft, or unpredictable.

For example, catching a ball requires more than simply moving an arm to the correct position. The robot must coordinate its movements, respond to the ball's motion, and manage the forces generated during contact.

Similarly, hitting a moving ball with a bat requires accurate timing, coordinated arm movements, and the ability to produce enough force at the right moment.

These challenges are part of a broader robotics problem known as dexterous manipulation — the ability to control and handle objects through complex, coordinated movements.

Andrew S. Morgan, the first author of the AthenaZero research paper, explained that object manipulation has remained a difficult problem since the early days of robotics.

The research team wanted to explore whether a robot designed with mechanical properties closer to those of a human could perform these tasks more effectively.

Meet AthenaZero: A Robot Built for Agile Movement

AthenaZero is a dual-arm robotic system featuring a central body, two robotic arms, hands, motors, and gears. Its structure is designed to support coordinated movements while allowing the arms to respond to forces from their surroundings.

Although the structural sections of its arms are rigid, their joints can yield to external forces in a controlled way. This allows the robot to respond more naturally when its hands make contact with an object or surface.

The design is important because physical interaction is not always about applying maximum force. Sometimes, a robot needs to move gently, absorb an impact, or adjust its position when an object behaves differently than expected.

AthenaZero was designed to offer a wider range of force control, allowing it to perform both delicate and forceful movements.

According to Morgan, conventional industrial robots are often optimized for strength and precision, but these characteristics can make them less suitable for gentle physical interaction.

AthenaZero takes a different approach by focusing on the mechanical properties that influence how quickly and easily its arms can move and respond to forces.

The Secret Behind Its Agility: Low Inertia

One of the most important ideas behind AthenaZero is its low-inertia design.

In simple terms, inertia is an object's resistance to changes in its motion. A heavier object generally requires more force to accelerate, slow down, or change direction than a lighter object with a similar shape and mass distribution.

This principle also applies to robotic arms.

When a robot has high inertia, it may need more effort to accelerate or stop its movements quickly. This can make it harder to react to fast-moving objects or adjust its motion during physical contact.

By carefully designing the robot's mechanical structure and actuators, the researchers aimed to create arms that could change their movements more effectively.

Actuators are the components that produce movement in a robot, typically by converting electrical energy into mechanical force.

AthenaZero uses custom actuators to help define its mechanical properties. These components play an important role in determining how the robot generates force and responds to external interactions.

The researchers believe that combining advances in electric motor design with improvements in computing could help make this approach increasingly practical.

However, low inertia alone does not guarantee humanlike performance. A robot must also coordinate its movements, control its forces, and respond appropriately to changing conditions.

AthenaZero explores how mechanical design can support these abilities.

Why Did Researchers Choose Baseball?

To evaluate AthenaZero, the researchers selected three baseball-inspired activities: throwing a ball, catching a ball, and hitting a ball with a bat.

Baseball provides a useful test environment because it combines several challenging physical skills.

Throwing requires the robot to coordinate its arms and generate controlled motion. Catching requires it to respond to an approaching object and manage the forces created during contact.

Hitting a ball with a bat introduces another challenge: the robot must coordinate its movements to make contact at the appropriate moment.

These activities also help researchers examine how the robot performs when both arms need to work together.

Unlike a simple pick-and-place operation, baseball-inspired tasks require movements that change quickly and involve physical interaction with objects.

In their initial experiments, AthenaZero successfully performed the three tasks, demonstrating that its mechanical design could support coordinated and agile arm movements.

The results suggest that building robots for dynamic manipulation may require a different approach from simply making conventional industrial robots stronger or more precise.

The experiments do not mean AthenaZero can compete with professional baseball players or handle every unpredictable situation. Instead, the tasks serve as practical demonstrations of its ability to perform complex movements.

How Could This Technology Change the Future of Robotics?

The potential applications of this research extend well beyond sports.

Robots that can interact with objects more naturally could eventually become useful in manufacturing, logistics, healthcare, and other environments where objects vary in shape, weight, and texture.

In factories, for example, a robot with better force control could handle delicate components while still performing demanding physical tasks.

In warehouses, improved coordination and responsiveness could help robotic systems manage objects that move or shift unexpectedly.

The technology could also contribute to the development of more capable service robots. Such machines may need to pick up fragile items, manipulate tools, open containers, or assist people with everyday activities.

Another potential benefit is safer physical interaction. Robots that can better regulate the forces they apply may be more suitable for environments where people work nearby.

Nevertheless, these applications remain future possibilities rather than capabilities already demonstrated by AthenaZero. Additional testing would be needed to establish how reliably the system performs across different objects and real-world conditions.

What Comes Next for AthenaZero?

The research team plans to continue improving AthenaZero's movement and object-manipulation abilities.

The baseball experiments are an initial step toward a much larger goal: developing robots capable of performing dexterous tasks in dynamic environments.

Future work will focus on expanding the robot's capabilities and exploring more challenging forms of manipulation.

One major challenge will be ensuring that the robot can adapt to objects with different physical properties and respond reliably when conditions change unexpectedly.

Researchers will also need to determine how well the approach performs outside controlled experiments and how it compares with other robotic designs.

Morgan and his colleagues see AthenaZero as an early demonstration of a low-inertia robotic design strategy, rather than a finished solution to the broader problem of robotic manipulation.

Their work highlights an important idea in robotics: improving a machine's physical design may be just as important as improving its software and intelligence.

Conclusion

AthenaZero demonstrates how rethinking the mechanical design of robotic arms could help machines perform faster, more coordinated movements.

By combining a low-inertia structure, custom actuators, and joints that respond to external forces, the robot successfully completed baseball-inspired throwing, catching, and batting tasks.

Although it is still an experimental research system, the project offers a promising direction for developing robots that can interact with the physical world more naturally.

The ultimate goal is not simply to build robots that are stronger than humans, but to create machines that can adjust their movements, control their force, and handle objects with greater flexibility.

If this approach continues to advance, tomorrow's robots could become much more capable of performing the complex physical tasks that humans accomplish every day.

Research reference: Andrew S. Morgan et al., “AthenaZero: A low-inertia, bimanual robot for dynamic manipulation,” Science Robotics (2026). DOI: 10.1126/scirobotics.aee1868.