Event Camera Odometry Enables Real-Time Rover Navigation on Mars

A new research paper published on arXiv (2605.27661) presents a real-time asynchronous monocular odometry system designed specifically for planetary exploration rovers. Led by Benat Inigo and colleagues, the team addresses a critical challenge: rovers operate under strict computational limits yet must navigate unpredictable terrains with extreme lighting conditions (e.g., harsh shadows on Mars). Traditional cameras struggle with high dynamic range (HDR) and require high bandwidth, but event cameras solve this by reporting only pixel-wise brightness changes asynchronously, achieving microsecond temporal resolution. This drastically reduces data volume while preserving robustness.

The proposed approach uses an Error-State Kalman Filter (ESKF) to fuse the asynchronous event stream with the rover's motion model. The state is updated continuously using tracked positions from RATE, a real-time asynchronous feature tracker. Preliminary results show the system can estimate 6-DOF ego-motion reliably even in low-light and high-contrast scenarios—common on planetary surfaces. This design not only improves navigation speed but also cuts power and bandwidth usage, which are precious on space missions. The work is a step toward fully event-driven autonomy for rovers.