Ankush Dhawan

Robotics PhD @ Stanford

[Ankush Dhawan @ankushd5812]

@divyeshkrx This is going to be lit

[Ankush Dhawan @ankushd5812]

@matt__makes @Google I think people loved it!

[Ankush Dhawan @ankushd5812]

Pupper at SF tech week @Google for startups!

[Ankush Dhawan @ankushd5812]

@zhanpeng_he Very cool! Can the policies you train with this system still use the full 5 fingers of many dexterous hands even without recording data from the ring and pinky?

[Aiden Swann @SwannAiden]

Coming soon 🏌️

[Ankush Dhawan @ankushd5812]

Proud of this one! Geometric alignment between human and robot data feels like a key piece of making cheap, scalable human data more useful for robot learning. Great working with Julien, @MattStr10050461, and team!

[Matt Strong @MattStr10050461]

Introducing VisTacAlign 👁️👆📐: co-training dexterous policies on human demos collected with a different tactile sensor than the robot’s.

How we align vision and touch so cheap human data actually helps the robot! 🧵

[Matt Strong @MattStr10050461]

Introducing VisTacAlign 👁️👆📐: co-training dexterous policies on human demos collected with a different tactile sensor than the robot’s.

How we align vision and touch so cheap human data actually helps the robot! 🧵

[Robotics Papers @OWW]

VisTacAlign: Co-Training Dexterous Policies on Tactile Human and Robot Demonstrations

Julien Poffet, Matthew Strong, Ankush Dhawan, Baiyu Shi, Shalika Neelaveni, Yujia Yuan, Zhenan Bao, Monroe Kennedy
https://t.co/0ZM9JfkqP6 [𝚌𝚜.𝚁𝙾]

[Matt Strong @MattStr10050461]

Our lab has a new website! Check it out at https://t.co/z67Pb8BjJW.

[Ankush Dhawan @ankushd5812]

@govindchada @enact_labs Govind is back!

[Ankush Dhawan @ankushd5812]

@jamesw_stevens Super cool