Bio-Inspired Plastic Neural Networks for Zero-Shot Out-of-Distribution Generalization in Complex Animal-Inspired Robots
This work addresses the challenge of robust locomotion control for complex real robots, offering a solution for zero-shot out-of-distribution generalization, though it is incremental as it builds on existing Hebbian learning methods.
The researchers tackled the problem of neural networks failing catastrophically in out-of-distribution situations for robotics by improving Hebbian learning with weight normalization to prevent divergence, and demonstrated that this plastic network enables zero-shot sim-to-real adaptation and generalization to unseen conditions like uneven terrain and damage in real 18-DOF and 16-DOF animal-inspired robots.
Artificial neural networks can be used to solve a variety of robotic tasks. However, they risk failing catastrophically when faced with out-of-distribution (OOD) situations. Several approaches have employed a type of synaptic plasticity known as Hebbian learning that can dynamically adjust weights based on local neural activities. Research has shown that synaptic plasticity can make policies more robust and help them adapt to unforeseen changes in the environment. However, networks augmented with Hebbian learning can lead to weight divergence, resulting in network instability. Furthermore, such Hebbian networks have not yet been applied to solve legged locomotion in complex real robots with many degrees of freedom. In this work, we improve the Hebbian network with a weight normalization mechanism for preventing weight divergence, analyze the principal components of the Hebbian's weights, and perform a thorough evaluation of network performance in locomotion control for real 18-DOF dung beetle-like and 16-DOF gecko-like robots. We find that the Hebbian-based plastic network can execute zero-shot sim-to-real adaptation locomotion and generalize to unseen conditions, such as uneven terrain and morphological damage.