TY - GEN
T1 - MAPPS: A Multi-Agent Pick-and-Place System for Efficient Robotic Sorting
AU - Kokkalisxsp, Konstantinos
AU - Konstantinidis, Fotios K.
AU - Koskinopoulou, Maria
AU - Tsimiklis, Georgios
AU - Amditis, Angelos
AU - Frangos, Panayiotis
PY - 2025/12/3
Y1 - 2025/12/3
N2 - This paper introduces MAPPS, a conceptual framework for multi-agent pick-and-place in robotic sorting. MAPPS integrates multi-modal sensing, real-time classification, and a decision-making module for dynamic task allocation. Two use cases illustrate the requirements of homogeneous and heterogeneous multi-agent systems: Construction and Demolition Waste (CDW) and Municipal Solid Waste (MSW) sorting. A two-robot demonstrator validates the concept and preliminary results show that simple scheduling cannot fully exploit system capabilities. To address this, the framework incorporates Multi-Agent Reinforcement Learning (MARL) for adaptive coordination and improved efficiency. MAPPS establishes a foundation for scalable, intelligent robotic sorting in industrial environments.
AB - This paper introduces MAPPS, a conceptual framework for multi-agent pick-and-place in robotic sorting. MAPPS integrates multi-modal sensing, real-time classification, and a decision-making module for dynamic task allocation. Two use cases illustrate the requirements of homogeneous and heterogeneous multi-agent systems: Construction and Demolition Waste (CDW) and Municipal Solid Waste (MSW) sorting. A two-robot demonstrator validates the concept and preliminary results show that simple scheduling cannot fully exploit system capabilities. To address this, the framework incorporates Multi-Agent Reinforcement Learning (MARL) for adaptive coordination and improved efficiency. MAPPS establishes a foundation for scalable, intelligent robotic sorting in industrial environments.
U2 - 10.1109/ist66504.2025.11268423
DO - 10.1109/ist66504.2025.11268423
M3 - Conference contribution
BT - 2025 IEEE International Conference on Imaging Systems and Techniques (IST)
PB - IEEE
ER -
