🚀 I’m excited to share a big milestone for Emancro!
We’ve successfully piloted our AI-powered robot at one of the leading US hospitals taking on one of the most critical and error-prone hospital logistics tasks: medication cabinet restocking.
At the core is GEDIA — our proprietary Generalizable Dexterous Intelligent Agent that gives our robots the ability to adapt and perform complex tasks at scale.
👉 Watch the robot in action: (piped.video/watch?v=QZbx89Hp…)
👉 Learn more about Emancro: (emancro.ai/)
And this is only the beginning 🤖. We’re already developing our next-gen robot to expand into medical supply distribution, lab sample logistics, food, and linen services — on track to automate the majority of logistics tasks in hospitals and beyond.
1
9
224
Frederik Ebert retweeted
We've updated Bridge Data with 33k robot demos, 8.8k autonomous rollouts, 21 environments, and a huge number of tasks! Check out the new Bridge Data website: rail-berkeley.github.io/brid…
The largest and most diverse public dataset of robot demos is getting bigger and bigger!
2
29
144
31,101
Please find more information here: sites.google.com/berkeley.ed…
2
Frederik Ebert retweeted
We developed a touch-sensing robotic "thumb," extending on the GelSight design, that can sense contacts on multiple sides, and use deep nets to learn how to insert a plug into an outlet.
w/ A. Padmanabha, F. Ebert, S. Tian, @RCalandra, @chelseabfinn
sites.google.com/berkeley.ed…
2
21
92
Frederik Ebert retweeted
Learning to grasp & insert a plug with only tactile feedback, using a bright new sensor based on the GelSight design.
arxiv.org/abs/2003.06965
w/ Akhil Padmanabha, @febert8888, Stephen Tian, @RCalandra, @svlevine
We developed a touch-sensing robotic "thumb," extending on the GelSight design, that can sense contacts on multiple sides, and use deep nets to learn how to insert a plug into an outlet.
w/ A. Padmanabha, F. Ebert, S. Tian, @RCalandra, @chelseabfinn
sites.google.com/berkeley.ed…
1
21
85
Frederik Ebert retweeted
RoboNet: a new large-scale dataset collected across multiple robots, labs, viewpoints, and objects, for studying generalization of predictive models and controllers across different robots!
arxiv.org/abs/1910.11215
robonet.wiki/
video:
piped.video/watch?v=qd-sBiKG…
68
249
Frederik Ebert retweeted
Tired of your robot learning from scratch? We introduce RoboNet: a dataset that enables fine-tuning to new views, new envs, & entirely new robot platforms.
robonet.wiki
arxiv.org/abs/1910.11215
w/ Dasari @febert8888 Tian @SurajNair_1 Bucher Schmeckpeper Singh @svlevine
1
85
331
Check out our work on "Manipulation by Feel: Touch-Based Control with Deep Predictive Models". bair.berkeley.edu/blog/2019/…
2
3
Check out our work on video-prediction based robotic manipulation:
Model-based RL, from pixels, controlling a robot and generalizing to new objects (clothing, toys, etc.). All trained with unsupervised interaction!
This blog post, and accompanying paper, summarize two years of our model-based RL research: bair.berkeley.edu/blog/2018/…