2D Restaurant Map Used for Simulations.

Robots, and particularly multi-robot systems, are becoming increasingly important in a number of domains due to their ability to operate more systematically than humans in controlled and semi-controlled environments. The restaurant industry is one area where multi-robot systems would have a significant impact. This project aimed to simulate a system of multi-robot waiters in a 2D restaurant who's layout was based on a real restaurant. A system of autonomous waiters performed order delivery and clean-up tasks in the restaurant. These tasks were allocated to the waiter robots while optimizing cost of allocation (e.g., total distance to be travelled). The navigation was accomplished with a global and local planner that utilized A* search while avoiding collisions with other waiters, patrons and obstacles in the environment. The goal was to maximize customer throughput and minimize task failures by combining near optimal robot navigation with task allocation. Task allocation was computed using the Hungarian algorithm. A number of experiments were performed to evaluate the performance of the Hungarian algorithm and to determine the robustness of the navigation. The experiments conducted varied different parameters such as the total number of waiters, the manner of task allocation, etc. and compared them to a random allocation baseline. Continuous task allocation with four waiters was found to maximize throughput. Using Hungarian algorithm optimized the allocation significantly compared to random allocation.

Videos showing example experiments are located below.

Full Project Report Found HERE .

Note: Project conducted with Akhil Devarakonda, Malavika Manoj, Vishnu Prem, and Venkat Varun Velpula.

Multi-robot restaurant system simulation where tasks were continuously allocated to the waiter robots. The waiters are the colored dots: olive, red, green, and blue; the black dots represent the patrons. Video shows 15 minutes of simulation at 10x speed.

Multi-robot restaurant system simulation where tasks were allocated to the waiter robots once at least two waiters had completed their tasks and returned to their waiting locations. The waiters are the colored dots: olive, red, green, and blue; the black dots represent the patrons. Video shows 15 minutes of simulation at 10x speed.