Saurabh Gupta

25 posts

Saurabh Gupta

Saurabh Gupta

@_saurabhg

Assistant Professor at @ECEIllinois. PhD from @Berkeley_EECS. BTech CS@IITD.

Champaign, IL Katılım Haziran 2011
238 Takip Edilen671 Takipçiler
Saurabh Gupta
Saurabh Gupta@_saurabhg·
Densifying sparse reward functions with our learned value functions speeds up learning. Furthermore, our learned value function with a low-level controller in 3D navigation is able to navigate to the optimal goal, whereas policy evaluation navigates to the sub-optimal goal.
Saurabh Gupta tweet media
English
1
0
0
0
Saurabh Gupta
Saurabh Gupta@_saurabhg·
Checkout our new work at DeepRL / OfflineRL Workshops at #NeurIPS: Learning Value Functions From Undirected State-Only Experience. We study how we can derive useful behavior from undirected state-only (i.e. w/o action labels) data. Website: matthewchang.github.io/latent_action_… See 🧵below.
English
2
2
10
0
Saurabh Gupta retweetledi
Devendra Chaplot
Devendra Chaplot@dchaplot·
If you are attending #NeurIPS2021, visit our poster on Wednesday, Dec 8 at 4.30pm PST! neurips.cc/virtual/2021/p…
Devendra Chaplot tweet media
Devendra Chaplot@dchaplot

Presenting SEAL: Self-supervised Embodied Active Learning! #NeurIPS2021 SEAL is a self-supervised framework to close the action-perception loop. It improves perception & action models by just moving in the physical world w/o any human supervision. devendrachaplot.github.io/projects/seal 1/N

English
2
3
20
0
Saurabh Gupta
Saurabh Gupta@_saurabhg·
Checkout SEAL: Self-supervised Embodied Active Learning at #NeurIPS2021 Object detectors trained on Internet data + Self-supervised interaction with 3D environments ➡️ Improved object detectors for robots.
Devendra Chaplot@dchaplot

Presenting SEAL: Self-supervised Embodied Active Learning! #NeurIPS2021 SEAL is a self-supervised framework to close the action-perception loop. It improves perception & action models by just moving in the physical world w/o any human supervision. devendrachaplot.github.io/projects/seal 1/N

English
0
2
15
0
hardmaru
hardmaru@hardmaru·
Tesla’s AI Day presentation feels more like a talk given at an ML conference than an investor presentation 👏🏻 youtu.be/j0z4FweCy4M
YouTube video
YouTube
hardmaru tweet media
English
19
132
1.2K
0
Saurabh Gupta
Saurabh Gupta@_saurabhg·
Check out our #RSS2021 paper on visual navigation in under-canopy agriculture environments! We can drive close to half a kilometer w/o intervention. 1.7x longer than previous best w/ LiDAR, while lowering sensing cost by 50x. Collaboration w/ @daslab_uiuc, lead by @ArunUIUC .
Arun Narenthiran@arunnarenvs

Super excited to share my first robotics paper! Check out our #RSS2021 paper on Learned Visual Navigation for Under-Canopy Agricultural Robots. PDF: arxiv.org/pdf/2107.02792… Webpage with data: ansivakumar.github.io/learned-visual… (videos at 5x)

English
1
3
39
0
Saurabh Gupta
Saurabh Gupta@_saurabhg·
Stop by our #NeurIPS poster on "learning navigation policies by watching Internet videos" today at 9pm PST. Think offline RL but on real data, and without labels for intervening actions. Attendee details: neurips.cc/virtual/2020/p… Project website: matthewchang.github.io/value-learning…
Saurabh Gupta@_saurabhg

Check out our #NeurIPS2020 paper on learning semantic cues for navigation by watching in-the-wild videos. Project page with video and code: matthewchang.github.io/value-learning… Paper: arxiv.org/abs/2006.10034 w/ Matthew Chang, Arjun Gupta

English
0
2
24
0
Saurabh Gupta
Saurabh Gupta@_saurabhg·
Checkout our ECCV20 paper on cross-video cycle consistency for video understanding. w/ S. Purushwalkam, T. Ye, A. Gupta Poster: - Tue 25 Aug - 14:00-16:00 UTC +1 - Wed 26 Aug - 00:00-02:00 UTC +1 PDF: ecva.net/papers/eccv_20…
GIF
English
0
1
11
0