Me, Myself, and AI · MIT Sloan Management Review

AI on Mars: NASA’s Vandi Verma

December 19, 2023·27 min·2 clips

What can we learn from AI on Mars with NASA's chief engineer Vandi Verma?

This episode features Vandi Verma, chief engineer at NASA's Jet Propulsion Laboratory, discussing the practical application of AI in space exploration. Hosts Sam Ransbotham and Shervin Khodabandeh explore how AI operates in the extreme environment of Mars. Verma's work involves developing and operating the autonomous systems for NASA's Martian rovers. Verma explains that AI on Mars primarily handles autonomous driving and robotic arm operations for the Perseverance and Curiosity rovers. She details how the rovers use computer vision to navigate the Martian terrain without real-time human guidance. The planning software, developed by her team, allows a rover to assess its surroundings and execute a multi-step science mission independently. A key challenge is the communication delay, which can be up to 20 minutes one-way, making Earth-based direct control impossible. Verma shares that the rovers have driven over 50 kilometers autonomously on Mars. She describes specific tools like the "Planetary Robotics Lab" used to simulate and test rover activities. The episode covers how these systems prioritize safety, such as avoiding hazardous slopes, while pursuing scientific objectives. One surprising insight is that the AI does not use deep learning, relying instead on more deterministic, verifiable algorithms for critical navigation. Verma reveals that a typical Martian "sol," or day, involves uploading a single, comprehensive plan for the rover to execute. She notes the software's ability to handle unexpected events, like a wheel slipping, by recalculating its path. The development process involves extensive testing in terrestrial analogs like the Chilean desert. Verma emphasizes that the goal is not full autonomy but reliable automation that extends human decision-making. A memorable point is her comparison of rover software to a "remote scientist" that can perform detailed, pre-planned investigations. The tone is conversational, educational, and filled with genuine enthusiasm from both the hosts and the guest. The style is interview-based, focusing on practical explanations of complex engineering challenges. Listeners fascinated by space technology, robotics, and real-world AI applications outside of business contexts would love this episode. Those seeking discussions on commercial AI trends or generative AI might find the focus too niche.

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Me, Myself, and AI