Event
InteractAgent: A Human-Scene Interaction System with Memory Reflection
1 reports · 1 sources
Senhe Zhang et al. propose InteractAgent, a multimodal large language model agent system for long-horizon human-scene interaction, which improves task success rate and reduces penetration rate on the UniHSI benchmark compared with a training-free baseline. The system completes planning, execution, observation, and reflection in a closed loop: a single MLLM serves simultaneously as planner and evaluator, a frozen MotionLLM serves as executor, and online observation and memory-enhanced reflection mechanisms are incorporated.
Report timeline
- InteractAgent: A Memory-Augmented Reflection Agent for Long-Horizon Human-Scene Interaction Animation
The paper proposes InteractAgent, a multimodal large language model agent system for long-horizon human-scene interaction that completes planning, execution, observation, and reflection in a closed loop, with a single MLLM serving as both planner and evaluator and a frozen MotionLLM as the executor.
arXiv cs.RO ·