German researchers develop tool to enhance robot movement

Researchers at the Technical University of Munich (TUM) have developed a groundbreaking tool that leverages the natural oscillation patterns of the human and animal body to enhance robotic movement.

This tool calculates energy-efficient motions. It was tested on BERT, a robotic dog, which successfully demonstrated improved movement dynamics, Caliber.Az reports, citing foreign media.

By exploiting the body’s natural oscillations and perfecting timing, BERT was able to perform faster, more fluid movements, surpassing traditional robotic methods in experiments focused on efficient gait control. The researchers explain that this innovation allows robots to move more fluidly, mimicking the effortless movements of humans and animals that occur without conscious thought. Humans excel at performing daily tasks using their upper limbs, effortlessly handling objects with complex designs, flexible materials, or unpredictable behavior. Studies show that humans can predict and navigate chaotic systems, likely by tuning into natural motion patterns through the central nervous system (CNS).

By stimulating the CNS’s resonant frequency, this ability helps synchronize multiple movements while minimizing effort. Humans instinctively align with a system's natural frequency to control motion in predictable ways, as shown in everyday activities like bouncing a ball or jumping on a trampoline. By forming internal models of an object’s dynamics, humans can predict its movements and apply control strategies that exploit the system's behavior—even if it sacrifices energy efficiency. While human sensitivity to resonance has been studied in simpler linear systems, understanding this behavior in more complex, nonlinear systems remains a challenge. 

Nonlinear systems, such as a double pendulum, display chaotic and unpredictable dynamics, but they also exhibit periodic motions known as nonlinear normal modes (NNMs). These modes, which resemble the eigenmodes of linear systems, vary with energy levels and don’t always pass through equilibrium. In earlier research, the team developed a computational tool to explore whether humans intuitively align with these NNMs. The tool predicts the intrinsic motions in nonlinear systems, providing insights into how humans excite and stabilize resonance even in complex scenarios. 

Now, a team led by Albu-Schäffer at the Technical University of Munich (TUM) has applied this tool to optimize movement in robots. By identifying natural oscillation patterns, the tool helps researchers calculate the most energy-efficient movements for robots. The tool was tested on BERT, a small, dog-like four-legged robot created by Albu-Schäffer at the German Aerospace Centre (DLR). 

The aim is to develop more efficient and adaptable legged movement for robots. Through the research, the team discovered six energy-efficient movement patterns for BERT, including walking, trotting, and hopping. These patterns align with the robot’s inherent oscillations, proving that robots can achieve efficient gaits by utilizing their natural rhythms. 

By Naila Huseynova