ESCOM Students Develop AI-Powered Quadruped Robot

“Xiinbal” learns to walk and collect environmental data using Artificial Intelligence.

Computer Systems Engineering students from the Escuela Superior de Cómputo (ESCOM) at the Instituto Politécnico Nacional (IPN) have developed “Xiinbal,” a quadruped robot that learns to walk using Artificial Intelligence (AI). The robot is designed to navigate urban environments while collecting data on temperature, air, and noise pollution.

“Xiinbal,” which means “walker” in the Mayan language, is also capable of mapping uneven terrain. In the future, it could be deployed in emergency and rescue operations, such as locating victims trapped in collapsed buildings following earthquakes.

The project, launched in August 2023 as part of the thesis work of Ryan Nathanael Cruz Barragán and Mauricio Emiliano Ruiz Alamilla, is coordinated by Dr. Roberto Eswart Zagal Flores, head of ESCOM’s Urban Data Lab, and research professor René Baltazar Jiménez Ruiz.

Dr. Zagal, who holds a PhD in Advanced Technology, emphasized that Xiinbal is an open-source, low-cost technology platform, with both hardware and software designs available in public repositories.

“The robot’s structure mimics that of a quadruped animal, which enhances its mobility. Using a LiDAR sensor, it performs 3D scanning to detect obstacles,” he explained.

Professor Baltazar Jiménez, who holds a Master’s in Mobile Computer Systems, added that the robot learned to walk through a type of AI called Reinforcement Learning. This technique enabled Xiinbal to replicate the movement patterns of real quadrupeds, such as dogs or leopards.

Weighing approximately one kilogram, the robot is equipped with four 4.3-centimeter limbs and is powered by a 12-volt battery. Its central control system is based on “ESP32” development boards, which receive and execute commands from a remote computer. The LiDAR sensor also enables Xiinbal to detect nearby people and animals as it scans the terrain.

Ryan Nathanael Cruz Barragán shared that developing the robot pushed him to learn and explore new areas of knowledge. He recalled that in its early trials, Xiinbal struggled to walk, frequently falling and breaking parts—but once it took its first steps, its performance improved exponentially.

Mauricio Emiliano Ruiz Alamilla noted that the project drew on diverse fields of knowledge, including physics (motors, weight, gravity, orientation, friction), 3D analytic geometry, linear algebra for neural networks, and electronics.

He encouraged future students to go beyond graduation requirements and explore scientific and technological areas where they can make meaningful contributions.