Project: #22
Coordinated Control and Task Allocation for Autonomous Mobile Robots in Warehouses using Multi-objective Multi-agent Reinforcement Learning
SHRIMAN KESHRI
With increasing efficiency needs in warehouses, this research proposes the use of multi-objective multi-agent reinforcement learning (MO-MARL) strategies to develop coordination and task allocation algorithms for autonomous mobile robots in warehouse environments.
The use of MO-MARL is motivated by the existence of competing objectives in warehouse operations, such as time, energy and safety. One of the challenges is the presence of uncertainties in the consumer demand, which may lead to changes in desired optimal placement of the inventory, which needs to be accommodated in the algorithms. Another challenge is the non-uniformity in the requirements of warehouses, especially in MSMEs across the industries. Therefore, the learning mechanism in the proposed approach can facilitate generalization in the developed algorithms. The proposed project can lead to improved operational efficiency in a variety of warehouse management problems with coordinated control of autonomous mobile robots.