Perseverance Rover Smashes Autonomous Driving Record on Mars
NASA's Perseverance rover has achieved a significant milestone in autonomous navigation on Mars, completing approximately 90 percent of its travels without human intervention as of October 2024. This performance drastically surpasses the Curiosity rover, which operated autonomously for only about 6.2 percent of its journey. The breakthrough is attributed to the Enhanced Autonomous Navigation (ENav) algorithm, detailed in a study published in IEEE Transactions on Field Robotics. ENav enables the rover to navigate complex, uncharted terrain by analyzing onboard images and evaluating roughly 1,700 potential paths within a six-meter range. Despite operating on radiation-hardened hardware with computing power comparable to a late-1990s iMac G3, the system efficiently prioritizes routes based on travel time and roughness. It employs a specialized collision-checking process called ACE only for top-ranked paths to conserve resources. While stationary obstacles provide predictability, the lack of high-resolution orbital maps presents challenges. NASA continues to explore advanced AI integrations, including tests with Anthropic’s models, to further enhance waypoint generation and automation capabilities for future Martian exploration missions.
Wire timeline
Perseverance Rover Smashes Autonomous Driving Record on Mars
NASA's Perseverance rover has achieved a significant milestone in autonomous navigation on Mars, completing approximately 90 percent of its travels without human intervention as of October 2024. This performance drastically surpasses the Curiosity rover, which operated autonomously for only about 6.2 percent of its journey. The breakthrough is attributed to the Enhanced Autonomous Navigation (ENav) algorithm, detailed in a study published in IEEE Transactions on Field Robotics. ENav enables the rover to navigate complex, uncharted terrain by analyzing onboard images and evaluating roughly 1,700 potential paths within a six-meter range. Despite operating on radiation-hardened hardware with computing power comparable to a late-1990s iMac G3, the system efficiently prioritizes routes based on travel time and roughness. It employs a specialized collision-checking process called ACE only for top-ranked paths to conserve resources. While stationary obstacles provide predictability, the lack of high-resolution orbital maps presents challenges. NASA continues to explore advanced AI integrations, including tests with Anthropic’s models, to further enhance waypoint generation and automation capabilities for future Martian exploration missions.
IEEE Spectrum