Hierarchical Multi-agent Reinforcement Learning for - Grenoble, France - NAVER LABS Europe

NAVER LABS Europe

Entreprise vérifiée

Grenoble, France

il y a 3 semaines

Posté par:

Sophie Dupont

beBee Recruiter

Description

NAVER LABS Europe's Action Group is focussed on enabling embodied agents to efficiently execute complex tasks and navigate in dynamic environments.

Within the Optimization With Learning (OWL) team of the Action group, the intern will focus on the optimization of a fleet of robots.

Those robots must perform a set of tasks associated with specific locations, and navigate a constrained environment, avoiding one another, to complete their tasks.

This problem is decomposed into a bi-level optimization problem. At the upper level, the tasks and necessary resources are assigned to robots.

At the lower level, a multi-agent path finding problem (MAPF) is solved to optimize the displacement of the robots, avoiding collision.

The objective of the internship is to propose a multi-agent reinforcement learning (MARL) approach to optimize the assignment and scheduling of the tasks, given paths produced by a MAPF algorithm.

Formalizing such a problem for MARL requires tackling off-beat actions [Qiu et al. 2022], i.e., asynchronously executed temporally varying stochastic actions stemming from the actual dynamics of the robots.

Similar problems are often addressed in the literature with a proxy measure and a centralized controller, but recent results suggest MARL as a possible efficient approach for this problem [Krnjaic et al.

2023].

[Krnjaic et al. 2023] Scalable Multi-Agent Reinforcement Learning for Warehouse Logistics with Robotic and Human Co-Workers. Aleksandar Krnjaic, Raul D. Steleac, Jonathan D. Thomas, Georgios Papoudakis, Lukas Schäfer, Andrew Wing Keung To, Kuan-Ho Lao, Murat Cubuktepe, Matthew Haley, Peter Börsting, Stefano V. Albrecht.

arXiv:
v2

[Qiu et al. 2022] Off-Beat Multi-Agent Reinforcement Learning. Wei Qiu, Weixun Wang, Rundong Wang, Bo An, Yujing Hu, Svetlana Obraztsova, Zinovi Rabinovich, Jianye Hao, Yingfeng Chen, Changjie Fan. AAMAS 2023 Extended Abstract.

arXiv:

Supervisors:
Vassilissa Lehoux and Tomi Silander

Required skills

Master or Ph.

D. student in machine learning and/or combinatorial optimization

Good development skills (Python is preferred), experience in machine learning frameworks
Knowledge in reinforcement learning, and if possible combinatorial optimization

Application instructions

Please note that applicants must be registered students at a university or other academic institution and that this establishment will need to sign an 'Internship Convention' with NAVER LABS Europe before the student is accepted.

About NAVER LABS

NAVER is the #1 Internet portal in Korea with activities that span a wide range of businesses including search, commerce, content, financial and cloud platforms.

NAVER LABS, co-located in Korea and France, is the organization dedicated to preparing NAVER's future. NAVER LABS Europe is located in a spectacular setting in Grenoble, in the heart of the French Alps.

Scientists at NAVER LABS Europe are empowered to pursue long-term research problems that, if successful, can have significant impact and transform NAVER.

We take our ideas as far as research can to create the best technology of its kind.

Active participation in the academic community and collaborations with world-class public research groups are, among others, important tools to achieve these goals.

Teamwork, focus and persistence are important values for us.

NAVER LABS Europe is an equal opportunity employer.