Robotics in Harsh Environments: A Conversation with Dr. Mark Post
Host Galen Korosec sits down with Dr. Mark Post, Senior Lecturer in Robotics and Autonomous Systems at the University of York, to talk about where robotics stands today and where it's headed. The two first worked together on Galen's undergraduate modular robotics project, so the episode doubles as a catch-up between former supervisor and student. Mark's work centres on robots that can adapt and survive in harsh, unpredictable environments such as space, underwater, nuclear sites, and increasingly farms, all with minimal human intervention. The conversation moves from his current space-warehouse project to the deeper challenge of building machines that can reason, handle uncertainty, and earn human trust. Guest Dr. Mark Post, Senior Lecturer in Robotics and Autonomous Systems, University of York (UK), researching modular, reconfigurable, and self-managing robotic systems for space and other extreme environments. What's covered (03:06) How "new space" has pushed modularity, interoperability, and standard interfaces to the forefront (04:00) Project StarFab, a ground demonstrator for an autonomous in-orbit "space warehouse" that could store, assemble, service, and refurbish satellites without human hands (07:19) Distributed decision-making, ROS 2, and why software and intelligence remain the missing pieces (10:13) Why large language models fall short for safety-critical space work, and the case for fact-based AI (13:15) Mark's research into semantics, ontologies, and probabilistic reasoning for real-world uncertainty (19:00) The shift from "perfect" machines to probabilistic ones, and the human trust that comes with it (24:46) Why space is an ideal proving ground for autonomy that later benefits warehouses and farms (28:35) The biggest challenge ahead: the "mental inertia" of designers and decades-old computing paradigms (34:19) The coming value of verified facts and "chains of trust" for autonomous systems Key takeaways Space is moving from one-off, single-purpose builds toward modular, serviceable systems that can be reconfigured in orbit. For safety-critical autonomy, Mark argues we need a new kind of AI built on verifiable facts rather than language-model associations. Progress depends less on raw technology and more on a willingness to break old paradigms. "Please try to push new things forward and don't keep relying on old technologies all the time, because otherwise we'll never make progress." — Dr. Mark Post Further reading For the technical detail behind the space-warehouse concept, see Mark's co-authored StarFab paper from the 2024 International Astronautical Congress, on the York Research Database: https://pure.york.ac.uk/portal/en/publications/starfab-concept-of-operations-and-preliminary-system-definition-f/
