Algorithms Beyond the Horizon
Deep dives into GNC algorithms, sensor fusion, and autonomous systems for orbital and deep-space missions.
Algorithms Beyond the Horizon
Deep dives into GNC algorithms, sensor fusion, and autonomous systems for orbital and deep-space missions.
Latest from the Lab
- Beyond Reconstruction: Zero-Shot Anomaly Detection of Telemetry with TimeRCD
Space telemetry analysis is shifting from reconstruction-based models like TadGAN to Zero-Shot Anomaly Detection. Traditional generative models struggle in deep space, suffering from overfitting and failing to identify unprecedented operational faults. The TimeRCD framework introduces a paradigm shift via the Relative Context Discrepancy (RCD) approach. Instead of reconstructing nominal data, TimeRCD measures vector discrepancies between local and global temporal contexts using pre-trained Foundation Models. Validated on NASA’s SMAP and MSL datasets, this zero-shot architecture drastically cuts false positives, reduces computational loads on Rad-Hard processors, and paves the way for true onboard spacecraft autonomy. - Generative AI Rover Navigation: Technical Analysis of Claude Integration on Mars
This technical report analyzes the pivotal event of Sol 1707-1709, where Perseverance navigated Mars guided by the Claude Code AI agent. For the first time, an LLM was integrated into the critical path, generating executable Rover Markup Language (RML) code. This “Ground-in-the-Loop” paradigm transcends traditional AutoNav limitations through semantic terrain analysis and terrestrial computing power. Rigorously validated via Digital Twin simulation, the AI successfully directed 456 meters of autonomous driving. This milestone demonstrates Generative AI’s capacity to control physical assets in critical environments, redefining space exploration standards. - Automotive vs. Space LiDAR: Why Your Car’s Sensor Can’t Survive Mars
Why doesn’t NASA use standard self-driving car sensors on Mars Rovers? It seems cost-effective, but the reality is harsh. This article dives into the Automotive vs. Space LiDAR debate, analyzing why terrestrial tech—like Luminar’s Iris—fails off-planet. We break down the engineering nightmares of space: deadly cosmic radiation, vacuum outgassing, and extreme thermal cycles that shatter standard electronics. By comparing commercial efficiency against NASA’s reliability, we explain why “good enough” for Earth is fatal on Mars. Plus, discover the surprising exception where a $129 drone part actually made it to the Red Planet.
Meet the founder
Hi, I’m Tommaso. Aerospace Engineering student & Rover enthusiast. Currently designing Mars rover prototypes at Politecnico di Milano with the PoliSpace team. I created Beyond The Joystick to bridge the gap between complex GNC algorithms and real-world robotics. From embedded systems to autonomous navigation, I’m here to explore how machines understand the unknown.
