Fereshteh Sadeghi

Title: Fereshteh Sadeghi: Innovator in Visual Servoing Technology

Introduction

Fereshteh Sadeghi is a prominent inventor based in Mountain View, CA (US). She has made significant contributions to the field of robotics, particularly in the area of visual servoing. Her innovative work has led to the development of a unique patent that enhances the capabilities of robotic systems.

Latest Patents

Fereshteh Sadeghi holds a patent titled "Viewpoint invariant visual servoing of robot end effector using recurrent neural network." This patent focuses on training and utilizing a recurrent neural network model for visual servoing of a robot's end effector. The model is designed to generate action predictions at various time steps, guiding the end effector toward a target object. Notably, the model is viewpoint invariant, allowing it to be applied across different robots with varying vision components. It can also adapt to significant changes in the viewpoint of a single robot's vision system. The training process involves a large quantity of simulated data, supplemented by a smaller amount of real training data.

Career Highlights

Fereshteh Sadeghi is currently employed at Google Inc., where she continues to push the boundaries of robotic technology. Her work is characterized by a strong emphasis on integrating advanced machine learning techniques into practical applications.

Collaborations

Throughout her career, Fereshteh has collaborated with notable colleagues, including Alexander Toshev and Sergey Levine. These partnerships have further enriched her research and development efforts in the field of robotics.

Conclusion

Fereshteh Sadeghi's contributions to visual servoing technology exemplify her innovative spirit and dedication to advancing robotics. Her patent and ongoing work at Google Inc. highlight her role as a leading inventor in this dynamic field.