Spiking Neural Networks for Continuous Control via End-to-End Model-Based Learning
This work addresses the challenge of using SNNs for high-dimensional motor tasks in robotics, representing an incremental advance in spiking network applications.
The paper tackled the problem of applying spiking neural networks (SNNs) to continuous motor control by training fully spiking architectures end-to-end for robotic arm control, achieving stable training and accurate torque control in tasks like a 6-DOF simulated robot.
Despite recent progress in training spiking neural networks (SNNs) for classification, their application to continuous motor control remains limited. Here, we demonstrate that fully spiking architectures can be trained end-to-end to control robotic arms with multiple degrees of freedom in continuous environments. Our predictive-control framework combines Leaky Integrate-and-Fire dynamics with surrogate gradients, jointly optimizing a forward model for dynamics prediction and a policy network for goal-directed action. We evaluate this approach on both a planar 2D reaching task and a simulated 6-DOF Franka Emika Panda robot. Results show that SNNs can achieve stable training and accurate torque control, establishing their viability for high-dimensional motor tasks. An extensive ablation study highlights the role of initialization, learnable time constants, and regularization in shaping training dynamics. We conclude that while stable and effective control can be achieved, recurrent spiking networks remain highly sensitive to hyperparameter settings, underscoring the importance of principled design choices.