This academic article introduces a novel framework called D4GR (Dialogue for Goal Recognition) aimed at improving the interaction of assistive robots with humans, especially those with cognitive impairments. The D4GR framework enhances the robot’s ability to recognize human plans and goals from their actions, even when these actions are sub-optimal. This is achieved by using language feedback and world state information in a hierarchical task model. The article demonstrates that D4GR performs better than traditional Hierarchical Task Networks (HTN) in goal and plan accuracy. The research also underscores the potential of D4GR in real-world scenarios, such as helping people with cognitive impairments accomplish everyday tasks.
Publication date: 5 Oct 2023
Project Page: Not Provided
Paper: https://arxiv.org/pdf/2310.02462
