#safeRL - Twitter Hashtag

9 days ago

Check out our new result on Constraint RL! The algorithms are simple variations of SAC and MPO and make training with reward hierarchies and with constraints stable. Thanks to the amazing team! #ReinforcementLearning #Robotics #SafeRL #Hyfydy #DeepLearning #LearningWithMuscles

Leonard T. Franz @LeonardFranz1

9 days ago

What if constrained RL did not need to "spend a safety budget"? We introduce our new principled constrained RL framework: "Stochastic Decision Horizons". Watch unpromising futures fade away for our musculoskeletal humanoid👇 #ReinforcementLearning #Robotics #SafeRL #Hyfydy

4

21

8

6

3K

0

18

1

4

2K

Leonard T. Franz @LeonardFranz1

9 days ago

What if constrained RL did not need to "spend a safety budget"? We introduce our new principled constrained RL framework: "Stochastic Decision Horizons". Watch unpromising futures fade away for our musculoskeletal humanoid👇 #ReinforcementLearning #Robotics #SafeRL #Hyfydy

4

21

8

6

3K

Manan Tayal @tayalmanan28

11 months ago

🚨 Our paper "Semi-supervised Safe Visuomotor Policy Synthesis using Barrier Certificates" has been accepted to @IEEECDC2025! We propose a semi-supervised framework to learn safe policies from visuomotor using CBFs. 📄 https://t.co/3OH049drfh #CDC2025 #SafeRL #Robotics #AI

tayalmanan28's tweet photo. 🚨 Our paper "Semi-supervised Safe Visuomotor Policy Synthesis using Barrier Certificates" has been accepted to @IEEECDC2025!

We propose a semi-supervised framework to learn safe policies from visuomotor using CBFs.

📄 https://t.co/3OH049drfh

#CDC2025 #SafeRL #Robotics #AI https://t.co/Oppiyulf9Q

0

11

1

0

271

Fei Miao 苗菲 @fei_fay

about 1 year ago

#ICRA #MARL #Robotics #SafeRL Our group will present two papers at ICRA 2025, Atlanta! On May 19 Monday, we are presenting YOLO-LLM: You Only LLM Once for Multi-agent Reinforcement Learning, https://t.co/JVO7Cf0C9o. On Wednesday, -3:50, room 303, session title: planning for autonomous racing, we will present the paper: "Safety Guaranteed Robust Multi-Agent Reinforcement Learning with Hierarchical Control for Connected and Automated Vehicles", https://t.co/RJv6X8urGh with the poster session followed the presentation.

0

6

0

209

Journal of Machine Learning Research @JmlrOrg

over 1 year ago

'OmniSafe: An Infrastructure for Accelerating Safe Reinforcement Learning Research', by Jiaming Ji et al. https://t.co/nD1P4HN3pS #saferl #safety #unsafe

0

7

0

1

789

Ming Jin

@MingJin_AI

almost 2 years ago

Dive deeper into "Safe RL for Smart Grid Control & Operations" with our tutorial slides. Open access for all: https://t.co/zGP0zxHT2F #SmartGrid #ReinforcementLearning #ControlSystems #AISafety #SafeRL @Safe_RL_Lab

0

5

2

0

221

Ming Jin

@MingJin_AI

almost 2 years ago

Excited to share our expertise in AI for power systems at the #SmartGridComm Conference at Oslo, Norway! Leading 2 tutorials & presenting 3 talks. #SmartGrid #Cybersecurity #SafeRL #ReinforcementLearning #LLM #Optimization @VT_ECE @SanghaniCtrVT

MingJin_AI's tweet photo. Excited to share our expertise in AI for power systems at the #SmartGridComm Conference at Oslo, Norway! Leading 2 tutorials & presenting 3 talks. #SmartGrid #Cybersecurity #SafeRL #ReinforcementLearning #LLM #Optimization @VT_ECE @SanghaniCtrVT https://t.co/sT8K9SyP8U

0

6

2

0

817

Ming Jin

@MingJin_AI

almost 2 years ago

Excited to present a #SafeRL tutorial at @IJCAIconf with my collaborator, Shangding Gu! 🧠🛡️ We'll share latest research in the field on building AI that's not just smart, but safe. Link to slides: https://t.co/yEnDDkoSrk

1

8

4

2

2K

Haitham Bou Ammar

@hbouammar

about 2 years ago

🌟 Excited to share our new research on Safe Reinforcement Learning using the Constraint Manifold! Our method ensures safety during the entire learning process in real-world robotics. #AI #Robotics #SafeRL Traditional RL methods often fail to guarantee safety during learning, leading to potential damage. Our approach ensures safety by exploring only within safe action spaces. #AIResearch #MachineLearning Key Contributions🛠️ Introducing the Constraint Manifold concept to represent safe robot configurations, using differential geometry to construct safe action spaces. #Robotics #DifferentialGeometry 🤖 Presenting ATACOM (Acting on the Tangent Space of the Constraint Manifold), a novel method for safe exploration and learning in robotics. #SafeRL #AI Theoretical Foundations📚 Our method is backed by strong theoretical analysis, using LaSalle’s Invariance Principle to prove safety guarantees. #AITheory #MachineLearning Algorithm Development💡 Practical techniques for implementing ATACOM include slack variable dynamics, smooth varying tangent space basis computation, and drift clipping. #AIAlgorithms 2D Collision Avoidance🚗 Our method shows improved safety and learning efficiency in 2D environments with static and dynamic obstacles. #Robotics #SafeExploration Robot Air Hockey🏒 Validating our method in a high-dimensional task like Robot Air Hockey, demonstrating scalability and robustness against model mismatches. #Robotics #AIApplications Empirical Validation✅ Experiments in both simulated and real-world environments confirm the effectiveness of our method in handling complex constraints. #AIValidation #MachineLearning Conclusion🏁 ATACOM provides a robust solution for safe reinforcement learning in robotics, ensuring that robots can learn and adapt safely in real-world environments. #FutureOfAI #SafeRL Future Work🔮 We're excited about the potential of ATACOM for various real-world applications in robotics. Stay tuned for more updates! #AIInnovation #Robotics Acknowledgments🙏 Thanks to our amazing team and collaborators for their contributions to this research. Special thanks to our funding sources for their support. #TeamWork #Research

hbouammar's tweet photo. 🌟 Excited to share our new research on Safe Reinforcement Learning using the Constraint Manifold! Our method ensures safety during the entire learning process in real-world robotics. #AI #Robotics #SafeRL

Traditional RL methods often fail to guarantee safety during learning, leading to potential damage. Our approach ensures safety by exploring only within safe action spaces. #AIResearch #MachineLearning
Key Contributions🛠️ Introducing the Constraint Manifold concept to represent safe robot configurations, using differential geometry to construct safe action spaces. #Robotics #DifferentialGeometry

🤖 Presenting ATACOM (Acting on the Tangent Space of the Constraint Manifold), a novel method for safe exploration and learning in robotics. #SafeRL #AI
Theoretical Foundations📚 Our method is backed by strong theoretical analysis, using LaSalle’s Invariance Principle to prove safety guarantees. #AITheory #MachineLearning

Algorithm Development💡 Practical techniques for implementing ATACOM include slack variable dynamics, smooth varying tangent space basis computation, and drift clipping. #AIAlgorithms
2D Collision Avoidance🚗 Our method shows improved safety and learning efficiency in 2D environments with static and dynamic obstacles. #Robotics #SafeExploration

Robot Air Hockey🏒 Validating our method in a high-dimensional task like Robot Air Hockey, demonstrating scalability and robustness against model mismatches. #Robotics #AIApplications

Empirical Validation✅ Experiments in both simulated and real-world environments confirm the effectiveness of our method in handling complex constraints. #AIValidation #MachineLearning

Conclusion🏁 ATACOM provides a robust solution for safe reinforcement learning in robotics, ensuring that robots can learn and adapt safely in real-world environments. #FutureOfAI #SafeRL

Future Work🔮 We're excited about the potential of ATACOM for various real-world applications in robotics. Stay tuned for more updates! #AIInnovation #Robotics

Acknowledgments🙏 Thanks to our amazing team and collaborators for their contributions to this research. Special thanks to our funding sources for their support. #TeamWork #Research

1

37

10

16

3K

Fei Miao 苗菲 @fei_fay

over 2 years ago

@fei_fay #MARL #SafeRL Our innovative approach employs a safe actor-critic algorithm with new techniques like the truncated Q-function and safe action mapping, ensuring optimal decision-making in Connected Autonomous Vehicles (CAVs).

0

1

0

137

CMU Safe AI Lab @cmusafeai

over 2 years ago

🚀 Don’t forget DSRL (Datasets for Safe RL)! 📷 With 38 datasets under safety constraint information and consistent API with D4RL, DSRL is your go-to for offline safe learning exploration and research. Check it out at https://t.co/KB3Zh1JWPM now!#Dataset #SafeRL #AISafety (4/4)

0

3

0

179

CMU Safe AI Lab @cmusafeai

over 2 years ago

🚀 Excited to announce our latest research on #SafeRL, with three high-quality packages released! We introduce a benchmarking suite tailored for offline safe learning challenges. Explore more details at https://t.co/ZhUsTyd8Hc #AI #RL #AISafety (1/4)

cmusafeai's tweet photo. 🚀 Excited to announce our latest research on #SafeRL, with three high-quality packages released! We introduce a benchmarking suite tailored for offline safe learning challenges. Explore more details at https://t.co/ZhUsTyd8Hc #AI #RL #AISafety (1/4) https://t.co/CB8Cj4RFbh

1

25

6

2

2K

CMU Safe AI Lab @cmusafeai

about 3 years ago

Is safe reinforcement learning *truly* safe? 🧐 Our ICLR 2023 paper unveils surprising vulnerabilities in trained safe RL policies! 🚨 Even tiny state perturbations can trigger unsafe behaviors. #ICLR2023 #SafeRL #ReinforcementLearning #Robustness #AdversarialAttacks [1/3]

cmusafeai's tweet photo. Is safe reinforcement learning *truly* safe? 🧐 Our ICLR 2023 paper unveils surprising vulnerabilities in trained safe RL policies! 🚨 Even tiny state perturbations can trigger unsafe behaviors. #ICLR2023 #SafeRL #ReinforcementLearning #Robustness #AdversarialAttacks [1/3] https://t.co/BkKiMmDK2x

1

9

4

1

796

Davide Tateo @davide_tateo

over 3 years ago

This is an amazing work from @liu_puze! The ATACOM approach is a very effective way to deploy #RL algorithms to real #robots. Stay tuned, soon exciting new developments! @ias_tudarmstadt #RobotLearning #SafeRL

Georgia Chalvatzaki

@GeorgiaChal

over 3 years ago

Very happy for the 3 accepted papers at #ICRA2023 for @CS_TUDarmstadt ! (1/3) Safe RL of dynamic high-dimensional robotic tasks: navigation, manipulation, interaction. @liu_puze @SnehalJauhri @davide_tateo @Jan_R_Peters https://t.co/5bs9rMWyGC

1

117

19

21

18K

0

6

2

0

577

Yali Du

@yalidux

over 3 years ago

The video recordings for the workshops have been uploaded! 🤩🤩🤩 Welcome to check link below for the Workshop on Safe Reinforcement Learning Theory and its Applications 👏👏👏 https://t.co/7YMNLq9ANT #RL #safeRL

Yali Du

@yalidux

almost 4 years ago

1/2 Check out the 1st International Workshop on Safe RL Theory and its Applications at 21 Sep, UK (Online). We are excited to have invited the renowned researchers Prof. Sergey Levine @svlevine and Ding Zhao @zhao__ding as keynote speakers https://t.co/7YMNLq9ANT

1

8

0

1

0

5

1

0

RL-CONFORM IROS Workshop @RLCONFORM

almost 4 years ago

Thanks to all authors for their submissions! We will now start the reviewing process and send out notifications Sept 15. Stay tuned! #safeRL #robustRL #robotics #humanrobotinteraction #IROS2022

0

3

1

0

Jason Yang @JasonYa73655465

almost 4 years ago

In Vienna for #SafeRL workshop at #IJCAI2022

1

8

0

Floris den Hengst @FlorisHengst

almost 4 years ago

In Vienna for the #IJCAI2022workshop on #saferl

0

9

1

0

Max Mowbray @MaxMowbray3

about 4 years ago

Had a great time presenting work conducted as part of #CAFE4DM with @DongdaZhang at #ESCAPE32 today. I’ll also be presenting work with @Pnosapp and @AntonioE89 on #SafeRL for online optimization at the poster session tomorrow if interested! @UoMSciEng

MaxMowbray3's tweet photo. Had a great time presenting work conducted as part of #CAFE4DM with @DongdaZhang at #ESCAPE32 today. I’ll also be presenting work with @Pnosapp and @AntonioE89 on #SafeRL for online optimization at the poster session tomorrow if interested! @UoMSciEng https://t.co/xkWOnaOmJs

0

13

0

Yali Du

@yalidux

about 4 years ago

Check our recent review on safe reinforcement learning, along with an open-sourced suite of safe RL algorithms.#saferl Feedback and comments are warmly welcomed! [2205.10330] A Review of Safe Reinforcement Learning: Methods, Theory and Applications https://t.co/BCWL2n9g0T

2

6

2

0

Top Tweets for #safeRL

Last Seen Hashtags on Sotwe

Trends for you

Most Popular Users