Generated by All in One SEO v4.9.10, this is an llms.txt file, used by LLMs to index the site. # Christopher Thierauf, PhD Ocean Robotics and Autonomy Researcher ## Sitemaps - [XML Sitemap](https://www.cthierauf.com/sitemap.xml): Contains all public & indexable URLs for this website. ## Posts - [Updates](https://www.cthierauf.com/updates/) - Ongoing research, projects, and fieldwork - [What Matters in Designing Resilient Systems](https://www.cthierauf.com/publications/what-matters-in-designing-resilient-systems/) - Christopher Thierauf, Matthias Scheutz, IEEE ERAS 2025. How should our autonomous systems be designed around failure? We argue that we shouldn't design around failure as a special case at all: instead, failure should be treated as just another state to plan around, and we must instead interpret and integrate failure into our planning problem. When - [Bard College Seminar 2026](https://www.cthierauf.com/other/bard-college-seminar-2026/) - I presented on oceanographic technology and my work at Bard College in late March. If you're here from that presentation, you may find these links interesting for follow-on reading or reference. AUV Sentry is part of NDSF at WHOI. I'm part of the team of engineers that works on that system and helps deploy it - [Deployment and Development of a Cognitive Teleoreactive Framework for Deep Sea Autonomy.](https://www.cthierauf.com/publications/teleoreactive-cognitive-sentry-mission-executive/) - I show we can make deep-sea robots smart enough to act on their own with an at-sea deployment of AUV Sentry: the first step to capable deep-sea partners. - [Design of a Depth Triggered Mechanical Fuse.](https://www.cthierauf.com/publications/depth-triggered-mechanical-fuse/) - We describe a mechanical fuse, to make sure our deep-sea assets return to the surface even under the most catastrophic failure conditions. - [A Gentle Introduction to Symbolic Planning and Reasoning](https://www.cthierauf.com/other/a-gentle-introduction-to-symbolic-planning-and-reasoning/) - Symbolic planning, decision making, and logics made simple(r). I wrote a document introducing symbolic planning it in less formal terms. - [Robot Development and Path Planning for Indoor Ultraviolet Light Disinfection.](https://www.cthierauf.com/publications/path-planning-for-indoor-ultraviolet-light-disinfection/) - I describe a robot that can map its own environment, and then use this knowledge to plan a path that is provably capable of disinfecting a full region. - [“Do This Instead”—Robots That Adequately Respond to Corrected Instructions.](https://www.cthierauf.com/publications/robots-that-respond-to-correct-instructions/) - I present a method for robots that can handle self-corrections in human commands, and show through a human subjects experiment that it is the preferred method. - [Self-Debugging Robots: Fault recovery through reasoning and planning.](https://www.cthierauf.com/publications/self-debugging-robots/) - Can robots reason about what actions to take to better understand failure? I show a technique and demonstrate how this is possible. - [Toward Competent Robot Apprentices: Enabling Proactive Troubleshooting in Collaborative Robots.](https://www.cthierauf.com/publications/proactive-troubleshooting-in-collaborative-robots/) - We present a method for how can robots use dialog to explain, understand, and resolve their own failure. A human subjects experiment shows it is preferred. - [Fixing Symbolic Plans with Reinforcement Learning in Object-Based Action Spaces.](https://www.cthierauf.com/publications/fixing-symbolic-plans-with-rl-in-object-based-action-spaces/) - We can redesign the typical reinforcement learning pipeline to train faster and integrate with symbolic plans by training on the environment directly. - [ACuTE: Automatic Curriculum Transfer from Simple to Complex Environments](https://www.cthierauf.com/publications/acute-automatic-curriculum-transfer/) - Reinforcement Learning problems can be reduced into simpler steps that improve learning. We present an algorithm showing we can automate that on a real robot. - [Robots that Learn to Solve Symbolic Novelties with Self-Generated RL Simulations.](https://www.cthierauf.com/publications/solving-symbolic-novelties-with-rl-by-generating-simulations/) - We solve hard problems by imagining different outcomes in our head. How can robots create simulation environments to do the same? ## Pages - [Chris Thierauf, PhD](https://www.cthierauf.com/) - I'm a research engineer at WHOI studying how robots can become trusted partners for ocean exploration, while supporting science via AUV Sentry fieldwork. - [Chris Thierauf Research Papers -- Read Online](https://www.cthierauf.com/research/) - Full listing of Chris Thierauf's publications and ongoing research projects, with full text and summaries available. ## Categories - [Publications](https://www.cthierauf.com/category/publications/) - [Fieldwork](https://www.cthierauf.com/category/fieldwork/) - [Other](https://www.cthierauf.com/category/other/) ## Tags - [Mechanical](https://www.cthierauf.com/tag/mechanical/) - Mechanical design work. - [Autonomy](https://www.cthierauf.com/tag/autonomy/) - Autonomy projects, primarily long-horizon autonomy and task reasoning. - [DYNOS](https://www.cthierauf.com/tag/dynos/) - Anything pertaining to the DYNOS architecture for deep-sea robot autonomy. - [Reinforcement Learning](https://www.cthierauf.com/tag/reinforcement-learning/) - Posts about reinforcement learning. - [Resilience](https://www.cthierauf.com/tag/resilience/) - Posts about how robots can make decisions that improves overall task performance, even when failure occurs, or that can avoid the failure in the first place. - [Neurosymbolic](https://www.cthierauf.com/tag/neurosymbolic/) - Integrations between symbolic systems, like DYNOS-R or DIARC, and machine learning policies. - [HRI](https://www.cthierauf.com/tag/hri/) - Posts about HRI (Human-Robot Interaction): what behaviors can robots perform to make them more helpful partners? - [DIARC](https://www.cthierauf.com/tag/diarc/) - Posts involving the DIARC cognitive architecture from the Tufts Human-Robot Interaction lab.