Ravi Tejwani

As we aim to enable robots to engage socially with other agents, as much as we do as humans, there is a need for a rich theory of social interactions. We formalize this by extending Social MDPs where agents reason about the arbitrary functions of each others hidden rewards with different levels of reasoning. The extended Social MDPs encode five basic social interactions: cooperate, conflict, competition, coercion and exchange and can produce actions that are close to human judgements.

Preprint (Under review in ICRA, 2022)
Neural Information Processing Systems (NeurIPS) Workshop on Cooperative AI, 2021
Extended Social MDP Homepage

Just like humans understand the social interactions as we interact with each other there is a need for robots to reason the basic social interactions such as helping or hindering. We introduce, Social Markov Decision Processes(MDP), in which robots learn to reason about the physical and social goal of the other robot. Social MDPs allow specifying reward functions in terms of the estimated reward functions of other robots, modeling interactions such as helping or hindering another robot (by maximizing or minimizing the other robot reward) while balancing this with the actual physical goals of each agent.

Conference on Robot Learning (CoRL), 2021
International Conference on Intelligent Robots and Systems (IROS) Workshop on Cognitive HRI, 2021
International Conference on Robotics and Automation (ICRA) Workshop on Social Intelligence, 2021
Social MDP Homepage

We propose a novel alternative to backpropagation. This approach shifts the focus towards explicit propagation of neuronal activity by introducing noisy auxiliary variables which break the gradient chain into local, independent, layer-wise weight updates that can be done in a parallel manner. We provide theoretical convergence guarantees for a general class of online alternating optimization methods. Promising empirical results using multiple datasets and network architectures demonstrate that the new approach can perform on par with the state-of-art stochastic gradient descent (SGD) implementations of backprop and often learns faster initially, when only a small amount of data is available for training.

International Conference on Machine Learning (ICML)

Migratable AI is a system and method that I developed during my masters at MIT Media Lab. It allowed an embodied AI agent to migrate from its physical embodiment to another by maintaining the same persona and context of the dialog conversation. The context included the personal and non-personal utterances in different settings (public or private) of embodiment into which the agent migrated. The system was implemented in Robot Operating System (ROS). A dataset was collected from the dialog conversations between crowdsourced workers and further used to train the generative and information retrieval models. The system was validated with the human experiments through 2x2 between-subjects study on 72 users.

International Conference on Robot and Human Interactive Communication (RO-MAN), 2020
International Conference on Social Robotics (ICSR), 2020
Preprint (Under review)
Migratable AI Homepage