Meet CBI’s industry mentors: Jon Francis
Jul 14, 2025
CBI industry mentors Shalabh Jain, Jon Francis, Joao Semedo and Emily Ruppel with CBI president Alessandro Oltramari
With years of technical expertise and industry knowledge, four accomplished research associates from Bosch serve as industry mentors for the Carnegie Bosch Institute (CBI) fellows. These professionals bring years of applied research experience, innovation leadership, and domain-specific insight to the program, enriching the fellowship experience with a real-world lens. Their mentorship enhances the program’s academic rigor and entrepreneurial focus, ensuring that CBI fellows are equipped to translate emerging technologies into meaningful, real-world solutions.
Each mentor offers specialized expertise in one of four transformative areas driving the future of global industry: robotics, sustainability, artificial intelligence (AI), and safety and security. Beyond sharing technical know-how, they offer strategic guidance and career perspectives—helping fellows understand the broader ecosystem in which their research will operate and thrive.
We are pleased to introduce Jon Francis, CBI industry mentor focusing on the area of robotics. Francis is a senior research scientist in robot learning and multimodal machine learning at the Bosch Research and Technology Center right here in Pittsburgh, where he explores the intersection of machine intelligence, perception, and physical systems. We asked Francis some questions.
Where did you get your Ph.D.?
I earned my Ph.D. from the Language Technologies Institute in the School of Computer Science at Carnegie Mellon University; my dissertation is at the intersection of multimodal machine learning, embodied AI, robot learning, and neuro-symbolism.
What interested you in that topic and how did it/does it relate to what you do now?
I most value the chance to blend theoretical breakthroughs in machine learning, AI, and robotics with practical real-world applications. In the era of versatile large-capacity world and action models, developing autonomous systems that can reason about their surroundings, learn efficiently from human demonstrations, monitor their own progress towards a goal, and employ corrective actions to remediate their own suboptimal behavior has become increasingly possible.
My work pursues improvements in the intelligence and adaptability of autonomous systems that are deployed to diverse environments. On the theoretical side, I am considering the following paradigms:
- Robot trajectory retrieval/attribution for augmenting test-time policy learning, optimizing for data (re-)use
- Generalizing and adapting robot policies, online, via guidance from agentic frameworks (e.g., chain-of-thought reasoning, in-context learning, and code generation)
- Learning multimodal/multisensory object-centric representations for improved understanding of object contact physics, object affordances, and object surface properties in tabletop- and mobile manipulation settings
- Cross-embodiment transfer learning to scale methods across different robot platforms
On the application side, one of my topics is on learning robot policies that can automate challenging manufacturing tasks, such as dexterous manipulation of deformable objects for assembly; another topic looks at language-grounded robot decision-making and task-execution in home environments and intra-logistical factory/warehouse settings using foundation models.
What do you see as the future of your field?
The future of this field includes the deployment of ubiquitous autonomous systems that learn safe, robust, multimodal, and transferrable representations of the world and are therefore capable of acquiring, adapting, self-orchestrating, and improving their own skills/behavior over time—even when deployed to diverse unseen scenarios.
Why are you involved in CBI as an industry mentor?
I am excited to help shape CBI’s robotics research agenda (e.g., through the fellowship program, collaborations, and other funded initiatives) for two reasons, primarily because:
- I believe there is value in encouraging the research communities to tackle real-world problems and to be subject to real-world evaluation criteria and metrics
- I believe that resolving the challenges in the fields of robotics and embodied intelligence would allow us to make significant progress towards the fundamental goals in AI, e.g., general-purpose intelligence