Future Explained
Snake Robots Just Searched Venezuela's Earthquake Rubble. Explained: How 'Geometric Mechanics' Could Change Search and Rescue
Carnegie Mellon University's snake robots searched Venezuela's earthquake rubble in July 2026, the same team's second real disaster deployment since Mexico City in 2017.
CMU Biorobotics Lab team and rescue workers in Venezuela, July 2026. Photo: Carnegie Mellon University.
Key Points
- Two earthquakes struck northern Venezuela on 24 June 2026, killing more than 5,000 people; within a week, Carnegie Mellon University's Biorobotics Lab had a team and its snake robots on the ground in La Guaira.
- It was not the first deployment. The same lab sent an earlier version of the robot into a collapsed Mexico City building after the magnitude 7.1 earthquake of September 2017.
- Neither deployment found a survivor, but rescue teams used the robot's video feed to confirm several locations were empty, saving search time in structures with no one left to find.
- The underlying difficulty is mathematical as much as mechanical: a snake robot has far more joints than a conventional robot, and Howie Choset's lab has spent more than a decade working out how to coordinate them using geometric mechanics.
On 24 June 2026, two major earthquakes struck northern Venezuela, killing more than 5,000 people, injuring 16,800 and leaving 18,000 homeless. Within days, a team from Carnegie Mellon University's Robotics Institute was on a flight to Caracas, arriving in the coastal city of La Guaira within 24 hours of being asked to help. The team included Kimberly Elenberg, a retired US Army and Public Health Service nurse who has embedded with the Biorobotics Lab to bridge robotics and frontline medical response, alongside lab researchers Nico Zevallos-Roberts and Darwin Mick. Between 30 June and 3 July, they worked alongside the Venezuelan and Colombian Red Cross, Mexico's veteran Topos search and rescue brigade, and other international responders, threading snake-shaped robots into the gaps of collapsed buildings that no rescue dog or camera on a pole could reach.
How is a robot shaped like a snake actually useful in a collapsed building?
A conventional rescue robot has wheels or legs and needs a relatively clear path. A snake robot is built from many small motorised segments joined in a line, so it can lift its head to peer around a corner, flatten itself to slide under a slab, and push forward through a gap barely wider than its own body. That design lets the robot go where the rescue teams themselves cannot yet safely follow, into voids inside rubble that are too unstable or too narrow for a person.
How It Works
How It Works
Modular body.
The robot is built from many small motorised segments linked in a row, each one able to bend independently, rather than a single rigid chassis.
Distributed senses.
Cameras and lights are mounted along the body, letting the robot see and light its own path as it moves deeper into rubble, feeding video back through a tether to the operator.
Geometric mechanics.
Instead of commanding each joint separately, the control software coordinates every segment as a single travelling wave, the mathematical framework Carnegie Mellon's Biorobotics Lab has spent over a decade developing.
Multiple gaits.
That same wave-based approach lets the robot switch between movement patterns, sidewinding across loose debris, rolling, or inching forward through a narrow gap, depending on what the terrain demands.
Tethered control.
Power, video and commands travel through a physical cable rather than a wireless link, a deliberate choice given how unreliable radio signals are inside collapsed concrete structures.
Has this actually worked before?
The Biorobotics Lab's first real disaster deployment came nine years earlier, after a magnitude 7.1 earthquake hit Mexico City on 19 September 2017. A 2-inch-wide, 37-inch-long robot with 16 motorised segments made two passes through a collapsed apartment building, sending back video but finding no survivors. Matt Travers, a systems scientist on that team, said afterwards that "the robot performed well and the Mexican Red Cross workers with us said they would like to have a similar tool in the future." Choset called the result encouraging but said the robot still needed gas sensors and communication equipment before its next outing. In Venezuela in 2026, where national and international teams together rescued around 6,500 people from the rubble in the days after the earthquakes, the robot's outcome echoed 2017: no survivors located directly by the machine itself, but several sites confirmed as unoccupied, letting search teams redirect their limited hours and equipment toward buildings where people might still be alive. "This mission was a powerful reminder of both our shared humanity and the promise and challenges of deploying robotics during disasters," Choset said.
Why is coordinating a snake robot's movement so hard?
A robot arm or a wheeled rover has a handful of joints to control. Choset's robots have many more, what roboticists call degrees of freedom, and every additional joint multiplies the number of ways the machine's body can move without actually going anywhere useful. Simply telling each segment where to go, one at a time, does not scale: a 16-segment robot has too many possible combinations of joint angles for an operator, or even a conventional control algorithm, to search through in real time. That is the problem geometric mechanics was built to solve, and it is the reason the same underlying design now shows up far beyond search and rescue: snake robots from the same lab have inspected a defunct nuclear power plant, explored underwater environments, and been adapted into surgical tools that use ultrasound and force feedback to navigate blood vessels inside the human body.
What happens next?
Nine years separate the Mexico City and Venezuela deployments, and the robot has changed less than the mathematics behind it. Choset's own assessment after Mexico, that the tool needed better sensors before it was ready for wider use, still shapes what the lab prioritises: after Venezuela, the stated priorities are the same unresolved problems, detecting gas leaks and maintaining communication links inside structures that block radio signals and thick concrete. Every deployment also depends on a rescue team requesting the robot and a flight getting a small research lab's equipment into a disaster zone within days, a logistics chain that has held twice now without any guarantee it will hold wherever the next earthquake strikes. Whether a decade of refinement turns a robot that reliably rules out empty buildings into one that finds a survivor is the test the next earthquake will set, not this one.
Sources
- Carnegie Mellon University, "Snake Robots Support Earthquake Search and Rescue in Venezuela," July 2026.
- Carnegie Mellon University, "Snake Robots Aid Mexico Earthquake Search and Rescue," September 2017.
- CMU Robotics Institute, Biorobotics Lab, research agenda.
Institutions in this article: Carnegie Mellon University, Robotics Institute, Biorobotics Lab.
Frequently Asked Questions
Why were snake robots sent to Venezuela?
Two earthquakes struck northern Venezuela on 24 June 2026, killing more than 5,000 people. Carnegie Mellon University's Biorobotics Lab deployed snake robots to search collapsed buildings in La Guaira for survivors.
Did the snake robots find any survivors?
No survivors were located in either the 2026 Venezuela deployment or an earlier 2017 Mexico City deployment, but rescue teams used the robots' video feed to confirm several locations were empty, saving search time.
How does a snake robot move through rubble?
It is built from many motorised segments joined in a line, allowing it to lift its head around corners, flatten to slide under slabs, and push through gaps too narrow or unstable for a person or conventional robot.
Why is controlling a snake robot mathematically difficult?
It has far more joints, or degrees of freedom, than a conventional robot. Carnegie Mellon's Biorobotics Lab has spent more than a decade developing geometric mechanics, a mathematical framework for coordinating those joints into useful motion.
Has this snake robot technology been used for anything other than search and rescue?
Yes. The same underlying design has been adapted to inspect a defunct nuclear power plant, explore underwater environments, and guide surgical tools through blood vessels using ultrasound and force feedback.
What still needs to improve before the next deployment?
Gas leak detection and reliable communication inside structures that block radio signals, priorities Howie Choset identified after the 2017 Mexico deployment that still apply after Venezuela in 2026.
