A group of researchers has made a significant breakthrough in quadruped robotics , creating a robot dog model that emulates the adaptability of animals when moving through complex terrain. This innovation is based on a bio-inspired locomotion framework, which integrates gait strategies, procedural memory, and real-time adjustments.
A learning-based approach
The research team from the University of Leeds and University College London has developed a system that uses a deep reinforcement learning (DRL) model. This framework allows a quadruped robot to autonomously navigate uneven terrain, without the need for external sensors or prior training.
[post_relacionado]Unlike conventional systems that rely on a single gait strategy, the new model is capable of performing complex gait transitions and adapting to changing environmental conditions, improving its stability and resilience.
Robot dog adapts to animal movements
The secret of this innovation lies in the inspiration taken from animal biomechanics . Throughout evolution, quadruped mammals have developed a remarkable ability to adapt to different types of terrain through different gait modes, such as trotting, running or jumping.
This approach allows robots to mimic these capabilities, adapting their movement to the needs of the moment. The model developed by the researchers integrates three key attributes : advanced gait selection, procedural memory for rapid transition, and precise movement adjustments, offering a level of adaptability never seen before.
Current locomotion systems, although robust, cannot perform gait transitions fluidly, as they rely on predefined strategies that cannot adjust to out-of-the-ordinary conditions. In contrast, the system developed in this study employs an algorithm that enables autonomous gear shifting , similar to how animals do in the wild, making it a highly adaptable solution.
One step closer to autonomous robots
This bio-inspired locomotion framework has the potential to greatly improve the performance of robots in a variety of applications, from exploring difficult terrain to use in rescue tasks. Furthermore, by not relying on external sensors, the model presents a significant advantage by reducing the complexity and costs associated with traditional robotic systems .
Watch the robot walk on rough terrain and overcome obstacles. Source: Chengxu Zhou
The researchers also say that this development represents a breakthrough in robotics and also provides valuable insights into animal biomechanics, opening up new possibilities for the study and validation of biomechanical theories without the need for animal testing.
Follow us on social networks and don’t miss any of our publications!
YouTube LinkedIn Facebook Instagram X TikTok
Source: Arxiv
Photo: Chengxu Zhou