AI Controller Has Enabled Six-Legged Robot Locomotion
Researchers developed a system that allows robots to navigate uneven terrain and adapt to damage by mimicking insect gait.
Updated on Sept. 23, 2026 in Robotics
Live Poll
Should researchers prioritize biological mimicry when designing robots for complex or unpredictable environments?
Researchers from Tohoku University and VISTEC have developed an AI-based controller for a six-legged robot that successfully mimics the movement patterns of the stick insect Medauroidea extradentata. The research-stage system enables the robot to autonomously navigate unpredictable terrain and adapt to hardware failures.
Why it matters
Predefined walking patterns often fail on uneven surfaces or during mechanical failure, limiting the utility of autonomous machines. By using inverse reinforcement learning, this controller offers a way to generate flexible locomotion in environments where static programming is insufficient.
The system manages 18 individual leg joints using proximal policy optimization, an algorithm that improves stability during reinforcement learning. It outperformed static control methods by maintaining stable movement on uneven terrain despite being trained exclusively on flat surfaces.
The players
Tohoku University
A Japanese research university with a focus on robotics, materials science, and engineering.
VISTEC
The Vidyasirimedhi Institute of Science and Technology, a research-intensive institution focused on frontier science and engineering.
The details
The AI uses adversarial inverse reinforcement learning to derive the underlying reward structure of the stick insect. During locomotion, the controller continuously monitors body orientation and joint positions to adjust its movement in real-time. This allows the robot to reorganize its gait dynamically if one of its six legs is disabled, bypassing the limitations of rigid, pre-programmed motor sequences.
Timeline
September 23, 2026: Article publication date.
The Tech Race
This development moves beyond traditional rigid robotic programming by prioritizing biological adaptability. It represents a shift in the race to build autonomous systems capable of operating in unstructured defense and emergency response environments.
This technology is currently in the research stage and has not been integrated into commercial products. Future implementations may eventually influence the hardware design of machines intended for emergency response, infrastructure inspection, and defense applications.
The takeaway
The study demonstrates that AI can successfully mimic complex biological movement to increase the durability of robotic locomotion. Watch for future benchmarks evaluating how this controller performs on varying scales of hardware and in more extreme, unstructured environmental conditions.
Further reading
For more on developments in autonomous movement and design, visit Robotics.
Source note: This article includes information reported by IHLS.
Live Poll
Should researchers prioritize biological mimicry when designing robots for complex or unpredictable environments?
