Stuttgart – Prof. Dr. rer. nat. Marc Toussaint has begun his five-year tenure as a Max Planck Fellow at the Max Planck Institute for Intelligent Systems. Starting November 1st, he will be supervising a small research group at the Institute’s Stuttgart site. Toussaint is a Full Professor of Computer Science at the University of Stuttgart, where he has led the Machine Learning and Robotics Lab since 2012. He is one of 57 Fellows currently active at all Max Planck Institutes in Germany and abroad.
The Max Planck Fellow Program aims to promote cooperation between outstanding university professors and Max Planck researchers. By strengthening ties with universities, the program fosters cutting-edge research on both sides. During their initial five-year term, which can be extended, Max Planck Fellows typically set up a small lab at the host institute, funded by an annual budget of 100,000 Euros that is provided jointly by the hosting institute and the Max Planck Society.
Toussaint, who already has strong connections to researchers at the MPI-IS, is thrilled to have been nominated as a Max Planck Fellow. He sees it as a recognition of his work. “I am very pleased to be given this opportunity”, he says. “I think it’s necessary and beneficial for both the University of Stuttgart and the MPI-IS to have a closer cooperation, given our proximity and overlapping research interests. For me personally, I will benefit from the cutting-edge infrastructure provided here at the MPI-IS. I will receive an office, and I can hire several postdocs and Ph.D. students to form a team.”
“I am delighted that Marc Toussaint has been appointed as a Max Planck Fellow for the Stuttgart site of our institute”, says Katherine J. Kuchenbecker, Director of the Haptic Intelligence Department at the MPI-IS and currently Managing Director of the Stuttgart site. “Marc is a leading researcher in artificial intelligence and robotics, located only a kilometer away at the University of Stuttgart. I really want to strengthen the connection between our institutions. The Max Planck Fellow program is the perfect way to build that bridge.”
Toussaint’s group will be named the “Physical Reasoning and Manipulation Lab.” He plans to build up an experimental setup with two robotic arms that can interact with objects on a tabletop. Manipulating objects comes easily to a human, but the logic behind such actions is currently beyond what robots are able to do.
Algorithms that think creatively
Toussaint’s research focuses on the combination of decision theory and machine learning, motivated by applications in robotics. His goal is to develop learning systems that are able to reason about their own state of knowledge and decide which actions might yield the most informative future data, making them learn even better and eventually solve problems – a perfect fit for the research being conducted at the Max Planck Institute for Intelligent Systems. Further, Toussaint’s work is increasingly focused on real-world robotic systems and joint symbolic and geometric planning, including trajectory optimization and optimal control methods.
“I’m currently working on a project I want to focus my entire attention upon”, Toussaint explains. “I want to demonstrate that a robot can solve problems geometrically and physically. I want to be able to program a robot with a sense of what I call intuitive physics. I am working on programming algorithms that enable the robot to imagine what could physically happen. Imagine I ask a robot how to best build a treehouse in a forest. The robot would then give me several ideas about what material to use, different shapes of the treehouse, which tree would be best to use, and how to eventually build it. The robot would have an innate creativity and a basic understanding of physical processes. But which algorithms could enable the robot to be creative? That’s what I’m working on.”
“I saw Marc win the Best Paper Award at the prestigious Robotics: Science and Systems Conference in Pittsburgh this year. Only the best roboticists in the world win that prize”, says Kuchenbecker. “I was really impressed by how Marc is connecting ideas from science and intelligence, for instance how people solve problems logically, bringing these principles into robotic intelligence. This is an interdisciplinary cutting-edge research topic that he will be able to pursue here at the MPI-IS. It’s only by understanding physics, computation, and perception at a very deep level that one can give a robot physical intuition. Having Marc here at the institute will cross-pollinate with all the research groups we already have and increase the visibility of robotics and artificial intelligence in our region.”
Career not a straight line
Toussaint’s career is rather unusual. “It is no straight line”, Toussaint says with a smile. “I studied mathematics and then physics with a focus on gravitation theory, but I felt there were bigger open questions in other fields. So, I started working on neural networks as a doctoral student, wanting to understand intelligence.” In 2004, he received his Ph.D. from the Ruhr University Bochum. The title of his thesis was “The evolution of genetic representations and modular neural adaptation”. As a postdoc, Toussaint began shifting his research focus toward machine learning. He then came to do robotics. “For the past three or four years, I feel the robotics community is noticing me. My work is being acknowledged, and I feel I’m now an important part of the community. Becoming a Max Planck Fellow is testimony to that,” Toussaint says and explains what helps him succeed as a roboticist: “I really like to learn. Lots and lots and lots. That gives me a broad knowledge of many research areas in robotics and AI.”
Toussaint only recently was on sabbatical at Amazon’s research center in Berlin, where he headed the Machine Learning Robotics lab, and he was a visiting scholar at the MIT Computer Science and Artificial Intelligence Laboratory from 2017 to mid-2018. He has also been a Faculty Member of the International Max Planck Research School (IMPRS) for Intelligent Systems since the graduate school was set up. As a top robotics researcher at the University of Stuttgart, he is also part of Cyber Valley, the largest European research cooperation in the field of AI. Cyber Valley brings together the best scientists from around the world to accelerate their research in the Stuttgart-Tübingen region – Toussaint is one such exceptional talent, strongly contributing to the region’s leadership in AI research.
Find out more about Marc Toussaint at www.marc-toussaint.net and see him speak at his inaugural Max Planck Fellow lecture at the MPI-IS in Stuttgart on December 11th of this year. More details to follow.