Task partitioning in a robot swarm: object retrieval as a sequence of subtasks with direct object transfer

Area of Science:

• Robotics
• Artificial Intelligence
• Swarm Intelligence

Background:

• Task partitioning decomposes complex tasks into manageable subtasks for multiple workers.
• Sequential task execution requires inter-subtask communication and data transfer.
• Direct transfer involves immediate handover of outputs between sequential subtask performers.

Purpose of the Study:

• To investigate the benefits of task partitioning with direct transfer in swarm robotics.
• To evaluate the impact of task partitioning on mitigating dead reckoning errors during foraging.
• To demonstrate the feasibility of task partitioning with real robots in a foraging scenario.

Main Methods:

• A swarm of robots was employed for a foraging task, collecting objects from an unknown source and returning them to a home location.
• Two foraging strategies were compared: one without task partitioning and one utilizing sequential task partitioning with direct transfer.
• Dead reckoning was used for robot localization, with its inherent error accumulation being a key factor.

Main Results:

• Cooperation through task partitioning significantly limited the accumulation of dead reckoning errors.
• Task partitioning improved the swarm's capability in locating the object source.
• The implemented task partitioning strategy led to increased overall swarm performance during foraging.

Conclusions:

• Task partitioning with direct transfer is a viable strategy for enhancing swarm robotics performance.
• This approach effectively addresses limitations of dead reckoning in mobile robot systems.
• The study demonstrates practical benefits and feasibility through real-world robot implementation.