Carlos Santibanez and John Kern
Robotics, multi-agent systems, task allocation, Markov decision process
A task allocation algorithm for non-homogenous multi-agent systems based on human interaction is proposed. Through simulations, the algorithm has shown improvements in both operation time and energy usage over other task allocation algorithms. The interaction algorithm is modelled as a Markov decision process, which evaluates the capacity of a selected agent to perform a specific task while navigating in a dynamic environment. The proposed method aims to enhance performance in non-homogeneous groups compared to that of identical robots. Experimental results demonstrate that collaborative work is achievable and improved compared to other similar algorithms, even when human agents are involved. This work demonstrates that it is possible to collaboratively use different types of robots from diverse integrators, as is often seen in real-life workshops and industries.
Important Links:
Go Back
