Hybrid multi-robot task allocation using large language models

A method, comprising:

• generating, by a system comprising at least one processor and based on an output of a large language model (LLM), first assignment data representative of first allocations of respective first robots of a group of robots to respective tasks of a group of tasks, wherein the output of the LLM is generated based on LLM input data comprising capability information associated with the group of robots and task information associated with the group of tasks;
• transforming, by the system, the first assignment data to second assignment data using derivative-free optimization, wherein the second assignment data is representative of second allocations of respective second robots of the group of robots to the respective tasks of the group of tasks;
• and
• facilitating, by the system, performance of the respective tasks of the group of tasks by the respective second robots of the group of robots according to the second assignment data.

A non-transitory machine-readable medium comprising computer executable instructions that, when executed by at least one processor, facilitate performance of operations, the operations comprising:

• generating, based on an output of a large language model (LLM), first task allocations that assign respective first robots of a group of robots to respective tasks of a group of tasks, wherein the output of the LLM is generated based on input data comprising capability data associated with the group of robots and task data associated with the group of tasks;
• transforming the first task allocations to second task allocations via derivative-free optimization, wherein the second task allocations assign respective second robots of the group of robots to the respective tasks of the group of tasks;
• and
• causing the respective second robots of the group of robots to perform the respective tasks of the group of tasks according to the second task allocations.