Workflow from Scientific Research

Overview illustration of the general LRL process. Unlike the conventional multi-task approaches, where agents have simultaneous access to all tasks, an LRL agent can master tasks sequentially, one after another. Moreover, the agent should continually accumulate knowledge throughout the process. This concept emulates the human learning process.

DOI