@aprorok

CS Faculty @Cambridge_Uni

 Cambridge, United Kingdom
proroklab.org
Joined November 2011
  • Tweets122
  • Following220
  • Followers943
21Media
Our new paper crystallizes the role of heterogeneity in multi-agent reinforcement learning. We provide a new algorithm, and show that multi-robot systems with heterogeneous controllers boast higher resilience to observation noise: arxiv.org/abs/2301.07137
4
29
3,021
It was such a pleasure to visit Imperial again, thanks to @Ed__Johns for being a fabulous host, and thanks to the Imperial students for the vibrant Q&A after the session!
Edward Johns@Ed__Johns30 Jan 2024
It was my pleasure to host Amanda Prorok (@aprorok) last week, to hear her talk about how graph neural networks can learn intriguing, collaborative behaviours for multi-robot systems! You can watch a recording of her talk here: robot-learning.uk/seminar-se…. Enjoy!
4
940
We are super excited to soon be able to welcome you in Cambridge!!
Rika Antonova@contactrika19 Nov 2023
After a fruitful search for a faculty position in Machine Learning + Robotics, happy to share that I am joining the University of Cambridge. Interested in working together? Reach out to connect! (PhD applications due Dec 5th, MPhil - later in Spring) youtube.com/watch?v=LlCwHnp3…
2
875
Still a few days left to apply! If you're into robot learning, and like stuff that flies, check out our postdoc opportunity here at Cambridge: jobs.cam.ac.uk/job/43030/ @Cambridge_CL
1
9
1,309
Amanda Prorok retweeted
Andrew Davison
@AjdDavison23 Oct 2023
That is a good point and I was thinking something similar. Vision/Robotics/AI research is very much stuck in a groupthink bubble right now.
Soumith Chintala
@soumithchintala23 Oct 2023
Currently, we overfit to exploring AI techniques that work well on NVIDIA GPUs. With the AI Accelerator sanctions on China -- one interesting result might be that China forks silicon and explores a different idea-space in AI techniques than the rest of the world.
1
16
4,856
We have a Postdoc opening! Do you have deep learning skills and hands-on robot experience? Apply here to join our lab: jobs.cam.ac.uk/job/43030/
5
4
1,886
Join us next Monday at #ICRA2023 for our workshop on Multi-Robot Learning! We have a really neat line-up of speakers, spotlight presenters, and panelists! sites.google.com/cam.ac.uk/m… @cjdamato @angelaschoellig @ieee_ras_mrs @ieee_ras_icra

ICRA 2023 Workshop on Multi-Robot Learning

Although learning-based methods have become common-place for solving single-robot problems, they have only recently gained traction within the multi-robot domain. Practical multi-robot solutions lean...

sites.google.com
5
26
3,023
We're thinking about the benefits of diversity in multi-robot systems... Here's a cool movie that describes what heterogeneity can do for cooperative tasks: youtu.be/J81IVQEy-zw #HeterogeneousMARL #Robotics #GraphNeuralNetworks
2
18
2,473
This week at #ICLR2023, we are presenting POPGym, a comprehensive collection of partially observable RL problems and baseline models. Using POPGym, we execute a comparison across RL memory models 👇 arxiv.org/abs/2303.01859 github.com/proroklab/popgym
6
744
Check out our upcoming workshop on Multi-Robot Learning at #ICRA23! We have a super exciting line-up of speakers. Send us your submissions to feature in our spotlight and poster sessions! Deadline approaching: April 15. sites.google.com/cam.ac.uk/m… @cjdamato @angelaschoellig

ICRA 2023 Workshop on Multi-Robot Learning

Although learning-based methods have become common-place for solving single-robot problems, they have only recently gained traction within the multi-robot domain. Practical multi-robot solutions lean...

sites.google.com
8
1
37
5,503
Amanda Prorok retweeted
DeepAI
@DeepAI20 Jan 2023
Check out this awesome research paper by Matteo Bettini, @aprorok and others about Heterogeneous Multi-Robot Reinforcement Learning! 🤖🤖🤖 Read it now: deepai.org/publication/heter… #NeuralNetwork #ReinforcementLearning #Robotics #AI

DeepAI

Artificially intelligent tools for naturally creative humans.

deepai.org
1
3
1,229
At @Cambridge_CL's Prorok Lab, @aprorok and her team are working on finding practical methods to solve problems that require robots to coordinate with each other. #robotics #collectiveintelligence amazon.science/research-awar…

Amanda Prorok: Scaling new frontiers in multi-robotic research

The professor of collective intelligence and robotics at the University of Cambridge earned a 2019 Amazon Research Award for “Learning Explicit Communication for Multi-Robot Path Planning”.

amazon.science
3
1
24
We are hiring postdocs! A couple of keywords for the job: multi-agent systems, robots, GNNs, machine learning... If you're interested, apply here: jobs.cam.ac.uk/job/37669/
5
6
Too often, we focus on fixing the robot instead of the environment that it operates in. In this new paper, we do the latter, and show that subtle modifications to the robots’ surroundings make a real difference: arxiv.org/abs/2209.11279

Environment Optimization for Multi-Agent Navigation

Traditional approaches to the design of multi-agent navigation algorithms consider the environment as a fixed constraint, despite the obvious influence of spatial constraints on agents'...

arxiv.org
arXiv Daily
@Arxiv_Daily26 Sep 2022
Environment Optimization for Multi-Agent Navigation deepai.org/publication/envir… by Zhan Gao and @aprorok #ReinforcementLearning #ComputerScience

DeepAI

Artificially intelligent tools for naturally creative humans.

deepai.org
3
12
Thanks for the opportunity to speak at your summer school in Milan! What a vibrant crowd.
M2L school@M2lSchool16 Sep 2022
Thanks to the wonderful speakers of these last two days: @yayitsamyzhang Doina Precup @luisa_zintgraf @ja_kirkpatrick @aprorok @xbresson Chris Bishop, Ivan Nardini! What an amazing way to conclude the #M2Lschool 2022!
6
We are hiring! If you are passionate about all things flying, and have a knack for machine learning, then apply here: jobs.cam.ac.uk/job/36052/
10
2
38
Credit assignment among agents is really hard when you only get a global reward. In a new learning paradigm, agents use a #GNN to communicate key information that allocates appropriate credit, and promotes cooperative behavior: arxiv.org/abs/2205.13005

QGNN: Value Function Factorisation with Graph Neural Networks

In multi-agent reinforcement learning, the use of a global objective is a powerful tool for incentivising cooperation. Unfortunately, it is not sample-efficient to train individual agents with a...

arxiv.org
1
1
8
At #ICLR2022 we're presenting a new hybrid approach towards the TSP, that converges to optimal solutions, faster. Check it out: arxiv.org/abs/2110.05291

Graph Neural Network Guided Local Search for the Traveling...

Solutions to the Traveling Salesperson Problem (TSP) have practical applications to processes in transportation, logistics, and automation, yet must be computed with minimal delay to satisfy the...

arxiv.org
Cambridge University
@Cambridge_Uni26 Apr 2022
🧩 The 'Travelling Salesperson Problem' is one of the oldest puzzles in computer science. A new approach to solving it, developed at @Cambridge_CL, could improve efficiency in the transport and logistics sectors. @AProrok #ICLR2022 bit.ly/3Mq5ijU

Improved approach to the ‘Travelling Salesperson Problem’ could improve logistics and transport...

A notorious theoretical question that has puzzled researchers for 90 years, the Travelling Salesperson Problem also has real relevance to industry today. Essentially a question about how best to...

cam.ac.uk
1
7
Delighted to be one of Cambridge's 12 Pilkington Prize winners this year, recognizing outstanding teaching. Thanks to @Cambridge_CL for supporting my vision, and to a fabulous team of TAs for keeping the wheels turning! cst.cam.ac.uk/news/innovativ…
2
1
3
39
Looking forward to this!
KAUST CEMSE
@KAUST_CEMSE21 Feb 2022
#RobotoKAUST 2022 is approaching! Join us for the Robotics Technologies session on Monday, February 28, 14.30 - 18.45. Check the agenda and register cemse.kaust.edu.sa/robotokau… @aprorok @scoogan3 @bryanklow @gilleslubineau @tsiotrasP #cemseKAUST #Robotics
1
Load more