A larger presence of robotic systems and their impact on our society is dependent on their autonomous operation in uncertain environments, sensory requirements and interaction with human operators. This talk will present my research covering the topics of game theory, stochastic control, machine learning and some recent feedback control-enabled robotic applications. I will first present a pursuit-evasion differential game on the wall to highlight the importance of the game-associated value function and its analysis. By augmenting the well-known Hamilton-Jacobi-Isaacs equation with a novel rate-of-loss analysis of the value function, we show that the well-known differential game theory dilemma actually does not exist. Following the work on cracking the 60-year old game theory dilemma, I will cover the role of value function for switching, which is associated with robust decision making in stochastic environments. I will first introduce a Markov inequality-based rule for switching among time optimal controllers. With this, a larger stage is set for the use of statistical inference for switching among multiple controllers. This will be illustrated by several examples and by proposing hierarchical value-only based neural network training for switching among multiple controllers. The talk will be concluded with some recently developed feedback control-enabled sense of touch for soft robotic fingers without a force sensor and autonomous navigation for an electric tractor for weeding plants which won the Small Farm prize in the first-ever Farm Robotics Challenge (https://news.ucsc.edu/2023/06/farm-robotics-challenge.html).
Dejan Milutinović is the head of the Robotics and Control Laboratory at UCSC and a professor in the Department of Electrical and Computer Engineering, University of California Santa Cruz. He earned a Dipl.-Ing (1995) and Magister’s (1999) degrees in electrical engineering from the University of Belgrade, Serbia and a doctoral degree in electrical and computer engineering (2004) from Instituto Superior Tecnico, Lisbon, Portugal. From 1995 to 2000 he worked as a research engineer in the Automation and Control Division of Mihajlo Pupin Institute, Belgrade, Serbia. His doctoral thesis was the first runner-up for the best Ph.D. thesis of European Robotics in 2004 by EURON. He won the NRC award of the US Academies in 2008 and Hellman Fellowship in 2012. He has been with UCSC since 2009. Prof. Milutinović’s research interests are in the modeling and control of stochastic dynamical systems applied to robotics. His work is focused on fundamental problems related to the navigation of single and multiple autonomous vehicles, automation of surgical robots, soft robotic grippers and autonomous navigation for agriculture. He has served as an associate editor for multiple conferences and journals in robotics and control (IROS, ICRA, ACC, ICUAS, DSCC), ASME Journal for Dyn. Sys. Mes. and Control, IEEE Robotics and Automation Letters. He is currently a senior editor-at-large for the Journal of Intelligent and Robotic Systems and a senior member of the IEEE.
