About the Project
This project is a Mars rover simulation built in Isaac Sim featuring a neuromorphic navigation system designed for autonomous path planning in extraterrestrial environments. The simulation uses procedurally generated Mars-like terrain at 10 cm/pixel resolution based on real Mars topography data.
The core innovation is the Spiking Wavefront Planner (SWP), a neurobotic navigation system that uses Spiking Neural Networks (SNNs) and the E-prop learning rule to concurrently map environments and plan paths. This neuromorphic approach offers advantages in power efficiency and potentially speed when used with the appropriate hardware. It also opens up autonomous systems to use different analog/event-based sensing technologies.
I presented this work at the 2025 Human Augmentation Summit on August 23, 2025 in the MIT Media Lab as a member of the Augmentation Lab Residency Program. I also got to work with researchers at the MIT Media Lab and learn about research being done in autonomous systems for space exploration.
Resources
Demo Videos
Images
Mars Rover Pathfinding Simulation Interface