Reproduce “Closed-loop spiking control on a neuromorphic processor implemented on the iCub” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jingyue Zhao, Nicoletta Risi, Marco Monforte, Chiara Bartolozzi, Giacomo Indiveri, Elisa Donati
A spiking Proportional (P) controller was successfully implemented on the DYNAP-SE mixed-mode neuromorphic processor for closed-loop motor control of an iCub robot simulator, demonstrating robust performance in tracking tasks with optimized network structures against device mismatch. The controller achieved a lowest RMSE of 0.0228 in step response tasks, showcasing functional real-time control.
Key insights
- Spiking neural networks on mixed-mode neuromorphic hardware can effectively perform closed-loop motor control, demonstrating adaptive behavior in robotic tasks.
- Robustness against intrinsic hardware device mismatch and noisy inputs can be achieved through specific network architectural optimizations and redundancy, which is crucial for practical deployment.
- There exists a trade-off between control speed, sensitivity to network noise, and the smoothness/stability of motor trajectories, influenced by proportional gain and spike trace time constants.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.