Autonomous Reinforcement Learning Robot Control with Intel's Loihi 2 Neuromorphic Hardware
This work addresses the need for energy-efficient, real-time robotic control in space and terrestrial applications, though it is incremental as it builds on existing neuromorphic conversion methods and uses a previously validated controller.
The researchers tackled the problem of deploying reinforcement learning policies on neuromorphic hardware by converting trained artificial neural networks into spiking sigma-delta neural networks compatible with Intel's Loihi 2, demonstrating low-latency and energy-efficient inference for controlling the Astrobee robot in simulation.
We present an end-to-end pipeline for deploying reinforcement learning (RL) trained Artificial Neural Networks (ANNs) on neuromorphic hardware by converting them into spiking Sigma-Delta Neural Networks (SDNNs). We demonstrate that an ANN policy trained entirely in simulation can be transformed into an SDNN compatible with Intel's Loihi 2 architecture, enabling low-latency and energy-efficient inference. As a test case, we use an RL policy for controlling the Astrobee free-flying robot, similar to a previously hardware in space-validated controller. The policy, trained with Rectified Linear Units (ReLUs), is converted to an SDNN and deployed on Intel's Loihi 2, then evaluated in NVIDIA's Omniverse Isaac Lab simulation environment for closed-loop control of Astrobee's motion. We compare execution performance between GPU and Loihi 2. The results highlight the feasibility of using neuromorphic platforms for robotic control and establish a pathway toward energy-efficient, real-time neuromorphic computation in future space and terrestrial robotics applications.