|ANYmal Evolution and Locomotion
|Written by Harry Fairhead
|Sunday, 04 February 2024
From an early prototype in built at ETH Zurich in 2009, the ANYmal robot has evolved into a capable search and rescue robot. Two new videos reveal how sure-footed locomotion has been achieved.
This timeline from ANYbotics, the company founded in 2016 to industrialize the technology developed at the Robotics SystemsLab, shows how ALoF, the first legged robot at ETH has expanded its capabilities over the years to become a viable commercial products that in its latest incarnation Anymal X brings autonomous inspection to hazardous oil and gas and chemicals operations.
Research with ANYmal continues and this video from Fabian Jenelten and Junzhe He demonstrates ANYMal's search-and-rescue capabilities as the robot explores a disaster site training facility's challenging terrain.
In brief ANYmal's confdent locomation and ability for self recovery is due to combining trajectory optimization and reinforcement learning.
As the researchers explain:
Our approach utilizes a model-based planner to roll out a reference motion during training. A deep neural network policy is trained in simulation, aiming to track the optimized footholds. We evaluate the accuracy of our locomotion pipeline on sparse terrains, where pure data-driven methods are prone to fail. Furthermore, we demonstrate superior robustness in the presence of slippery or deformable ground when compared to model-based counterparts. Finally, we show that our proposed tracking controller generalizes across different trajectory optimization methods not seen during training. In conclusion, our work unites the predictive capabilities and optimality guarantees of online planning with the inherent robustness attributed to offline learning.
or email your comment to: firstname.lastname@example.org
|Last Updated ( Sunday, 04 February 2024 )