Wrapping Up 2025

Wrapping up 2025

Slowly getting ready for the New Year. I thought it would be time to reflect on the past year. It has been an interesting year with many activities. I was on half-time leave from UCSD April 2025 - March 2026 to be the inaugural Temasek Fellow. I’m still working with a number of venture capital funds. For the most part I disengaged from other companies such as Calibrate Ventures, … Still an advisor to Robo-Global which by now is owned by VettaFi and TMX.

Research is a fashion industry. It is interesting to see how different topics come and go. The last year in robotics it was all about machine learning models such as VLA and VLM. Obviously, it was also about humanoids. Doing real-world robotics to solve real industrial problems was much less sexy. It is easy to build a research system that has a 95% success rate. Unfortunately, such a system would fail an hour every day. Real success is when you can deploy a system that collaborates with people to solve real problems. Too many research projects are “college entertainment.” It is important to recognize that the ultimate benchmark is if people will pay for it for use in real applications. Robust.AI deployed robots for use with people in warehouses and operations centers. It was also an exercise in how do you go from a good prototype to a real product, and how do you build an associated support organization? Raising money for a real product is much harder than for an imaginary machine learning package. It is very challenging to raise money for a product with real revenue rather than a dream. Eventually the VC industry will have to do a reset.

In terms of research, we made real progress on ML models for platform generalization and methods for both locomotion and manipulation. The autonomous driving project saw real progress on the integration of language models, mapping, and prediction of the motion and intent of other road users, which is important when you drive on a college campus. Students have graduated and gone off to research labs and start-ups. The job market is far from trivial, and it is important for students to not just do well in classes but also build a portfolio of project examples that showcase the breadth and depth of their skill set. We have successful collaborations with Qualcomm, Nissan, Nvidia, IHI, SAAB, … Immensely grateful for their support of our research. We have a great pipeline of students. The ~10 PhD students lead the various projects, and we have 10+ MSc students that work with them. In turn, undergraduates are involved in model design, data acquisition, programming, and experimentation. Finally, we have a few high school students that participate in experimentation, data management, … So a full stack of people from high school to Ph.D. Truly grateful for all of their effort to make us a successful enterprise.

It has been a successful 2025 on the work front. I still see graduates regularly (thank you). With my wife, Marie, we have traveled the world. We saw family in Italy and the Georgia mountains. We still do food trips with friends (visited 8 Michelin stars this year).

Going into 2026, there is a lot of uncertainty. Can my students travel freely? Get the visa to be part of the broader science community? Will we have a new national (US) robotics program? What will the funding situation be like for research? How will the economy evolve in the new year? We cannot give up on a free press!

A Happy New Year to all of you, and thank you for all the engagements in the past year. May you all find growth, happiness, and peace.