Helping Hands Lab @ Northeastern

🤖 Robotic Manipulation | Reinforcement Learning | PI Rob Platt @RobotPlatt @KhouryCollege @Northeastern

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

SO(3)-equivariant policy learning in the RGB space needs much more exploration. Really like how this work streamlines the pipeline while preserving full symmetry. Amazing work from Boce. Highly recommend checking it out. See you at #NeurIPS2025!

[Boce Hu @boce_hu]

Closed-loop visuomotor control with wrist cameras is widely adopted and powerful, but leveraging full 3D symmetry from only RGB is still hard.

Introducing our #NeurIPS2025 Spotlight paper, Image-to-Sphere Policy (ISP), for equivariant policy learning from eye-in-hand RGB images.

[Boce Hu @boce_hu]

Closed-loop visuomotor control with wrist cameras is widely adopted and powerful, but leveraging full 3D symmetry from only RGB is still hard.

Introducing our #NeurIPS2025 Spotlight paper, Image-to-Sphere Policy (ISP), for equivariant policy learning from eye-in-hand RGB images.

[Dian Wang @Dian_Wang_]

Equivariant policy typically requires depth for symmetric representation, what if we only have a wrist-mounted RGB camera? ISP projects RGB image onto a sphere for 3D equivariant reasoning. @boce_hu and I will be at #NeurIPS2025 to present this paper, Fri. poster session 6 #2315!

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

Our new #ICML2025 paper, led by @ZhaoHaibo47588, presents a hierarchical equivariance architecture that enables multi-level sample efficiency in visuomotor policy learning. Check it out for more details!

[Haibo Zhao @ZhaoHaibo47588]

Excited to share our #ICML2025 paper, Hierarchical Equivariant Policy via Frame Transfer. Our Frame Transfer interface imposes high-level decision as a coordinate frame change in the low-level, boosting sim performance by 20%+ and enabling complex manipulation with 30 demos.

[Dian Wang @Dian_Wang_]

Honored to receive my first best paper nomination from @corl_conf! Had such a great time at #CoRL2024, huge thanks to the organizing committee and all my co-authors: Steve, David, @tarikkelestemur, @HaojieHuang13, @ZhaoHaibo47588, Mark, JW, @RobinSFWalters, @RobotPlatt!

[Snehal Jauhri @SnehalJauhri]

Perfect start to the #CoRL2024 week!
Was a pleasure organizing the NextGen Robot Learning Symposium at @TUDarmstadt with @firasalhafez @GeorgiaChal
Thanks to the speakers for the great talks! @YunzhuLiYZ
@NimaFazeli7 @Dian_Wang_ @HaojieHuang13 @Vikashplus @ehsanik @Oliver_Kroemer

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

Paper link: https://t.co/eTJd2HnNSv
Website: https://t.co/5zMSaGgJop
Coauthors: @Dian_Wang_, @BizaOndrej, @RubyFreax, @SeanLiu7081, @SaulBadman12, @RobotPlatt, @RobinSFWalters

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

#CoRL2024 IMAGINATION POLICY: Using Generative Point Cloud Models for Learning Manipulation Policies Led by @HaojieHuang13. A key-frame multi-task policy can generate key poses (imagine) and do manipulation precisely with sample efficiency. Presenting at Poster Session 4.

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

Paper link: https://t.co/GHdh0GeJVh
Website: https://t.co/Uwn6p9bBWM
Coauthors: @XupengZ, @BizaOndrej, Shuo Jiang, @LinfengZhaoZLF, @HaojieHuang13, @yuqi_Beijing, @RobotPlatt

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

#CoRL2024 ThinkGrasp: A Vision-Language System for Strategic Part Grasping in Clutter by @RubyFreax. A plug-and-play vision-language grasping system that uses GPT-4o’s advanced contextual reasoning for heavy clutter environment grasping strategies. Presenting at Poster Session 3.

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

Paper link: https://t.co/FAzp7UsgHz
Website: https://t.co/bcpRSOBjeg
Coauthors: Stephen Hart, David Surovik, @tarikkelestemur, @HaojieHuang13, @ZhaoHaibo47588, Mark Yeatman, Jiuguang Wang, @RobinSFWalters, @RobotPlatt

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

#CoRL2024 Equivariant Diffusion Policy Led by @Dian_Wang_. A sample efficient BC algorithm based on equi diffusion. It leverages symmetry to boost learning with 5x less training data and mastering complex tasks with <60 demos. Presenting at Oral Session 1 and Poster Session 2.

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

Presenting at Poster Session 2.
Paper link: https://t.co/6SmiY2tupo
Coauthors: Long Dinh Van The @cjdamato @RobotPlatt

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

#CoRL2024 Leveraging Mutual Information for Asymmetric Learning under Partial Observability led by @HaiNguy69482974
Addressing asymmetric learning under partial observability (state availability at training) by rewarding actions leads to histories that gain info about the state.

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

Paper link: https://t.co/TLeEHONTtv
Website: https://t.co/DH4QD2f9Ue
Coauthors: @XupengZ @Dian_Wang_ Zihao Dong @HaojieHuang13 @330781570 @RobinSFWalters @RobotPlatt

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

#CoRL2024 OrbitGrasp: SE(3)-Equivariant Grasp Learning Led by @boce_hu. Orbitgrasp maps each point in the cloud to a continuous grasp quality function using spherical harmonics. Our method outperforms all baselines across all settings and tasks. Presenting at Poster Session 1.

[Helping Hands Lab @ Northeastern @HelpingHandsLab]

Excited to share our #CoRL2024 paper: OrbitGrasp—a fully SE(3)-equivariant grasp detection method for 6-DoF grasping!
arxiv: https://t.co/eOi2DqYNEZ

[Boce Hu @boce_hu]

Grasp detection is crucial for robotic manipulation but remains challenging in SE(3).

We introduce our #CoRL2024 paper: OrbitGrasp, an SE(3)-equivariant grasp learning framework using spherical harmonics for 6-DoF grasp detection.

🌐 https://t.co/HzWP5OghY3

[Evangelos Chatzipantazis @EChatzipantazis]

Join us on Monday October. 14th at 2pm (UTC+4) in #IROS2024 Workshop on Equivariant Robotics.
A great lineup of keynote speakers will discuss how symmetry penetrates each and every subfield of robotics.
Website and Zoom Link: https://t.co/841TWqfvWq

[Linfeng Zhao @LinfengZhaoZLF]

We have released the YouTube recording of our #RSS2024 workshop on "Geometric and Algebraic Structure in Robot Learning"!
🎥Youtube: https://t.co/wCgGY9Bi3P
🏠Webpage: https://t.co/DBn6T2CNaa

[Robert Platt @RobotPlatt]

Instead of inferring a desired object pose directly, this method "imagines" a reconstruction of the entire scene in the target pose. Surprisingly, we find that this improves sample efficiency, even though we are inferring more information.