0
0
Advancements in Robotics: ANYmal Masters Parkour and Rough Terrain
ANYmal, a quadrupedal robot developed by researchers at ETH Zürich in Switzerland, has recently made significant strides in its capabilities. Originally designed for search-and-rescue operations in challenging environments, ANYmal has now been enhanced to perform basic parkour maneuvers, as detailed in a recent paper published in Science Robotics.
Evolution of ANYmal
The journey of ANYmal began in 2019 with an innovative approach to reinforcement learning, which was further refined over the years to include enhanced proprioception for improved movement and spatial awareness. In a notable demonstration, a trio of customized ANYmal robots were tested in terrain mimicking lunar and Martian surfaces, showcasing their adaptability and robustness.
The trio included the Scout model, equipped with RGB cameras for environmental surveys, the Scientist model with analytical tools for material identification, and the Hybrid model designed for versatile scientific measurements. Each robot had a specialized function while remaining flexible enough to assist one another in case of malfunctions.
Complexities of Parkour in Robotics
The latest milestone for ANYmal involved tackling the intricate task of parkour, a challenging feat for robots due to the dynamic and agile nature of the sport. The Swiss research team implemented a combination of machine learning and model-based control to enable ANYmal to perform dynamic movements, such as scaling obstacles, jumping across gaps, and adapting to rapidly changing environments.
By dividing the task into perception, locomotion, and navigation modules, the researchers enabled ANYmal to assess its surroundings, select appropriate movements, and navigate obstacles in real-time. Through machine learning and model-based control, the robot learned to recognize and negotiate gaps in rubble, showcasing its ability to adapt to unexpected scenarios.
Limitations and Future Prospects
While ANYmal has made significant progress in mastering parkour and rough terrain traversal, challenges remain in scaling the capabilities to diverse and unstructured environments. The team acknowledges the need for further testing and refinement to enhance the robot’s adaptability to a wider range of scenarios.
Despite these challenges, the advancements achieved with ANYmal represent a significant step forward in robotics, showcasing the potential for robots to navigate complex terrains with agility and precision. By striving to match the agility of free runners, researchers aim to push the boundaries of robotic capabilities and further refine the integration of perception, navigation, and actuation in robotic systems.
Overall, the development of ANYmal highlights the continuous evolution of robotics and the potential for robots to assist in various applications, including search-and-rescue operations and exploration in challenging environments.
Image/Photo credit: source url
