Meet X1: First multi-robot team used for rescue missions

Beyond Single-Purpose Bots: The Rise of Multimodal Robotic Teams

For years, the robotics industry has been divided into silos: robots that fly, robots that drive, and robots that walk. While each excels in its own domain, they all share a common weakness—they are defeated by environments that require more than one form of movement. A drone cannot carry heavy gear through a narrow hallway; a wheeled robot cannot scale a flight of stairs.

The emergence of systems like X1, a collaboration between Caltech’s Center for Autonomous Systems and Technologies (CAST) and the Technology Innovation Institute (TII), signals a shift toward “multimodal” robotics. Instead of building a better single machine, the trend is moving toward a “system of systems.”

By pairing a modified Unitree G1 humanoid with the M4 transforming drone, researchers have created a coordinated duo. This approach allows the humanoid to handle the heavy lifting and navigation of complex indoor terrains, while the drone provides aerial reconnaissance and rapid movement across open spaces.

Locomotion Plasticity: The Secret to All-Terrain Versatility

One of the most significant trends in autonomous systems is “locomotion plasticity”—the ability of a robot to switch its movement style based on the environment. The M4 drone exemplifies this by folding its rotors into wheels, allowing it to drive to conserve energy and then switch back to flight to overcome obstacles like water.

This flexibility is critical for real-world application. In a simulated emergency scenario, the X1 system demonstrated this by navigating tight hallways and doorways before the M4 drone launched to survey a campus pond. This ability to adapt ensures that a mission doesn’t grind to a halt just due to the fact that the terrain changes from a paved road to a flooded street.

Moving Away from Recorded Motion

Most humanoid robots traditionally rely on recorded human motion to walk or climb, which makes them struggle in unfamiliar or “messy” environments. However, the operate led by Aaron Ames at Caltech is pivoting toward physics-based models combined with machine learning.

This shift allows robots to infer patterns in data and adjust their balance in real-time. By focusing on the physics of movement rather than mimicking a human, these machines can maintain stability while carrying heavy payloads across varied terrain.

The Future of Autonomous First Response

The ultimate goal for multimodal systems is to serve as rapid first responders. In chaotic disaster zones where It’s too risky for humans to enter, a system like X1 can scout damaged buildings, localize itself without joystick control using lidar and range finders, and carry essential supplies.

To make this a reality, the industry is focusing on several key technical trends:

• Edge Computing: Engineers at TII are refining secure onboard computers and flight controllers, allowing drones to make swift decisions locally when networks fail.
• Safety-Critical Control: Developing methods that keep robot behavior safe even when sensors glitch or conditions become “noisy.”
• Hardware Security: Creating tools to guard robot hardware from interference during critical missions.

The integration of specialists—from Caltech’s focus on bipedal locomotion to Northeastern University’s tuning of morphing mechanisms—shows that the future of robotics is inherently collaborative.

Bridging the Gap Between Lab Demos and Public Trust

While the X1 has successfully navigated campus labs and libraries, the transition to real-world emergencies requires more than just technical prowess; it requires public and regulatory trust.

Future development is shifting toward “auditability.” In other words creating clear ways to audit a robot’s decisions and implementing emergency stops that humans can trigger instantly. Proving that these machines can perform long runs without crashes is the final hurdle before they move from the lab to the field.