Max Robotics — Global Service Robot Directory + US Certification

About

Deeksha Manjunath is a leading researcher at the intersection of robotics, machine learning, and real-world control systems. Her work is best known for pioneering scalable approaches that enable robots to learn from vast, diverse datasets—a critical step toward general-purpose robotics. She is the lead author of **RT-1: Robotics Transformer for Real-World Control at Scale** (2023, over 500 citations), a landmark paper that demonstrated how a Transformer-based model can be trained on millions of real-world episodes to perform hundreds of manipulation tasks with high success. This work has become foundational in the field of robot learning, showing that large-scale, task-agnostic data can be effectively transferred to downstream tasks. Manjunath also contributed to **Q-Transformer** (2023), which advances offline reinforcement learning by using autoregressive Q-functions to train multi-task policies from both human demonstrations and autonomously collected data. Her research has been instrumental in bridging the gap between large-scale machine learning and practical, real-world robotics. For her contributions, she has been recognized as a key figure in the movement toward foundation models for robotics, and her work continues to shape how robots are trained for complex, open-world environments.