Micro neural policies enable safe real-time robot control on tiny chips
Micro Neural Policies for Safe Real-Time Robotic Control
RoboticsArtificial Intelligence
Summary
Controlling robots safely and quickly is hard when using small, slow computer chips. This paper shows how tiny neural networks called Micro Neural Policies (MNP) can be trained to control robots like quadrotors and balancing poles without losing safety or performance. By combining special training methods and safety checks, the authors made these neural networks small enough to run fast on tiny chips with plenty of leftover computing power. They tested their approach in simulations and on real robots, proving that safety and speed can coexist on limited hardware.
What this means in practice
- •For embedded robotics engineers: Deploy small, verified neural controllers on microcontrollers to maintain safe and fast real-time control in resource-limited robots.
- •For industrial automation developers: Integrate lightweight neural policies to improve reliability and responsiveness of robotic arms and drones with constrained hardware.
Authors
Hongpeng Cao, Riccardo Curcio, Daniele Ottaviano, Marco Caccamo
Abstract
In this paper, we investigate the synthesis of Micro Neural Policies (MNP) to enable safe and robust real-time robotic control on computationally constrained embedded devices. We demonstrate that integrating Evolution Strategy (ES) and Statistical Model Checking (SMC)-based verification for policy search can drastically reduce neural network size without compromising safety and robustness. We conduct a large-scale training and evaluation of MNP on Cartpole and Quadrotor control tasks, varying control frequencies and network architectures. After validating these policies in simulation, we evaluate their deployability through zero-shot transfer to physical systems. Our experiments show that MNP can successfully achieve safe sim-to-real transfer without sacrificing control performance. We then show that the policies' memory footprint, ranging from 0.5 to 7.5 kB, allows deployment on microcontrollers, where they achieve real-time inference latency with under 25 ns of jitter while leaving the chip idle for over 97% of the time for additional workloads. This makes them a highly practical solution for severely resource-constrained robotic systems.