Help Me Explore: Minimal Social Interventions for Graph-Based Autotelic Agents
This work addresses skill acquisition in autonomous agents by integrating social guidance, which could enhance learning efficiency in robotics or AI domains, though it appears incremental by combining existing perspectives.
The paper tackles the problem of autonomous agents learning complex skills by proposing a social interaction protocol, Help Me Explore (HME), and a graph-based agent, GANGSTR, which together enable agents to master complex configurations like stacks of 5 blocks with minimal social interventions.
In the quest for autonomous agents learning open-ended repertoires of skills, most works take a Piagetian perspective: learning trajectories are the results of interactions between developmental agents and their physical environment. The Vygotskian perspective, on the other hand, emphasizes the centrality of the socio-cultural environment: higher cognitive functions emerge from transmissions of socio-cultural processes internalized by the agent. This paper argues that both perspectives could be coupled within the learning of autotelic agents to foster their skill acquisition. To this end, we make two contributions: 1) a novel social interaction protocol called Help Me Explore (HME), where autotelic agents can benefit from both individual and socially guided exploration. In social episodes, a social partner suggests goals at the frontier of the learning agent knowledge. In autotelic episodes, agents can either learn to master their own discovered goals or autonomously rehearse failed social goals; 2) GANGSTR, a graph-based autotelic agent for manipulation domains capable of decomposing goals into sequences of intermediate sub-goals. We show that when learning within HME, GANGSTR overcomes its individual learning limits by mastering the most complex configurations (e.g. stacks of 5 blocks) with only few social interventions.