Variable Friction Gripper on IEEE Spectrum

The design enables controlled within-hand sliding and rotation

In our collaboration work with Adam Spiers (Max Plank Institute) and Aaron Dollar (Yale University), we investigated the use of variable friction surfaces in within-hand manipulation. This human-inspired design makes it possible to slide and rotate the object within hand, allowing manipulation capabilities such as sliding-based gating and in-place rotation. Our work is covered by IEEE Spectrum Magazine, and draw a lot of interest from the manipulation community in the IROS conference.


Our Work Becomes ICRA 2018 Best Manipulation Paper Award Finalist

Learning Modes of Within-hand Manipulation

Our work that proposes a method to estimated modes of within-hand manipulation becomes best manipulation paper finalist in ICRA. The work uses a data driven approach to estimate whether the object is sliding or roling on the finger surfaces or whether it get close to the boundries of the workspace. Such modes are very crucial to know for conducting successful within-hand manipulation, but they are quite hard to determine with model-based approaches. Our method makes manipulation safer and much more reliable.


YCB Object and Model Set Project on the cover of 2017 Yale Engineering Magazine

"Grasping the future of robotics"

Aaron Dollar and I gave an interview for the annual Yale Engineering magazine about the Yale-CMU-Berkeley Object and Model Set Project. The article and the full magazine are available below.


Vision-based MPC improves in-hand manipulation efficiency

Model Predictive Control reduces the effects of modeling inaccuracies

Our latest results show that high performance in-hand manipulation can be achieved even with very rough process models. Combining visual feedback, system compliance and model predictive control framework allows us to minimize the effect of modeling inaccuracies and maintain efficiency. Our approach also removes the need for joint encoders and force sensors.


"A forgotten right of citizens: Technology"

A discussion on how to increase citizen participation in technological development

My article is published in a Turkish political magazine Ayrinti Dergi's special issue on citizenship (only available in Turkish).

Model-free active vision for robotics

An approach for viewpoint optimization using extremum seeking

We proposed a new methodology for optimizing the viewpoint of a robot's vision sensor without the need of an explicit objective function. The work enables new applications of active vision for robots operating in unstructured environments. We presented applications in object recognition and manipulation.