Great week at @ieee_ras_icra in Vienna. Presenting our work for underwater target tracking. https://t.co/KKD5mWLF67
As part of @DIGI4ECO and #TECTUGA@ICMCSIC
Parallelised Q-Networks go continuous! 🚀
We’re excited to introduce PQN extended to continuous-action control — no replay buffers, no target networks, just pure online Q-learning + a deterministic actor.
Now you can use PQN + MuJoCo Playground to train full robotic policies in just seconds or minutes 🤖 All running entirely on GPU
👉 Blog post: https://t.co/7lfLUccSGR
👾 Codebase: https://t.co/7S8z00IhWT
1/5
PQN has been accepted as a Spotlight at ICRL 2025 🎉
👉If didn't have time to check it out, take a look at our 5-minute blog covering PQN’s key features, plus a Colab demo with JAX & PyTorch implmentations https://t.co/wQobyj06qB
🔎 If you want to dive deeper, we break down TD instability issues and how PQN tackles them without replay buffers or target networks in this other blog post: https://t.co/LEnyNze1hl
See you in Singapore! 🇸🇬
⚪I OPEN CALL
🔬Vols impulsar la teva carrera investigadora? Guanya una de les beques INPhINIT de @FundlaCaixaCAT i vine a treballar a l'ICM, tenim una desena de projectes esperan-te!
⏳Deadline: 23 de gener de 2025
+ info: https://t.co/j7fsx0WaRF
@BecarisFLC#FundaciólaCaixa
⚪I OPEN CALL
🤖 Interested in marine robotics?
‼️ We offer a PhD position to work on the creation of advanced control strategies for Autonomous Underwater Vehicles, incorporating bioinspired design elements.
⌛30 November 2024
🔗https://t.co/HZhzgWAUHt
⚪I OPEN CALL
🪼¿Postdocs interesadxs cómo cambia la distribución de la biodiversidad debido al cambio global?
🤖¿Predocs amantes de la robótica submarina?
🚨En el ICM ofrecemos distintas plazas dentro del programa @Momentum_CSIC!
+ info aquí ➡️ https://t.co/xBcgW9Vwee
⚪I OPEN CALL
🪼¿Postdocs interesadxs cómo cambia la distribución de la biodiversidad debido al cambio global?
🤖¿Predocs amantes de la robótica submarina?
🚨En el ICM ofrecemos distintas plazas dentro del programa @Momentum_CSIC!
+ info aquí ➡️ https://t.co/xBcgW9Vwee
🚀 We're very excited to introduce Parallelised Q-Network (PQN), the result of an effort to bring Q-Learning into the world of pure-GPU training based on JAX!
What’s the issue? Pure-GPU training can accelerate RL by orders of magnitude. However, Q-Learning heavily relies on replay buffers and target networks, making training computationally slow and memory-intensive on GPUs. As a result, researchers often prefer PPO, leaving Q-Learning behind 🥲
Our solution? Eliminate Q-Learning's legacy components. PQN challenges the standard DQN paradigm by training a Q-Network without replay buffers or target networks: just online Q-Learning with vectorised exploration and network normalization (layer or batch norm). Despite its simplicity, PQN sets a new strong baseline in many single-agent and multi-agent scenarios. Check out the thread for more details 🔥
📄 Paper: https://t.co/tNz2ACyTOs
⚙️ Code: https://t.co/i4DuN9BEWR
A fantastic collaboration with @mattiefoxcs and @benjamin_ellis3, and the support of the amazing people at @FLAIR_Ox directed by @j_foerst.
Inspired by the groundbreaking work of @_chris_lu_ on compiling entire RL pipelines in GPU: https://t.co/4fA5ZAuMwC
Reinforcement learning can help autonomous surface vehicles to track target animals in the deep sea and could aid in many marine research endeavors.
Check out the 2023 Science #Robotics study on #WorldOceansDay: https://t.co/Kmsw2GpFnO
𝐓𝐡𝐞 𝐟𝐢𝐫𝐬𝐭 𝐝𝐢𝐬𝐬𝐞𝐦𝐢𝐧𝐚𝐭𝐢𝐨𝐧 𝐚𝐜𝐭𝐢𝐯𝐢𝐭𝐲 𝐨𝐟 𝐃𝐈𝐆𝐈𝟒𝐄𝐂𝐎 𝐩𝐫𝐨𝐣𝐞𝐜𝐭 𝐰𝐚𝐬 𝐚 𝐛𝐥𝐚𝐬𝐭!🚀
The project's coordinator, @ICMCSIC has participated in the “1st Korea-Spain Oceanographic Symposium” at Busan, Republic of Korea.
#digi4eco#cinea#eufunded
🤝Some days ago, we were at the Korea Institute of Ocean Science & Technology (KIOST) to build bridges to deepen the monitoring of MPAs through integrated technologies within the @DIGI4ECO#LifeECOREST#PLOME projects!
🎉Thank you all for attending, it was really enriching!
🔴I #ICMNEWS I 🔴
🤖 The European project @DIGI4ECO will improve marine conservation through digitization and underwater robotics.
🙌It will consider human pressure and climate change effects to design more effective management plans.
🔗https://t.co/FsbjbRofij
📷@OBSEAsarti