Task Planning for Robotic Disinfection Using Generative-Adversary-Trimodel (GAT)
Jiajie Ye, Yongji Sheng, Tianyu Liu, Ning Xi
AI summary
Problem
Robotic disinfection task planning in dynamic environments is a complex four-dimensional problem (interaction, logic, spatial, temporal) that requires expert knowledge, but traditional symbolic methods struggle with adaptability and learning-based methods lack strict spatio-temporal constraint handling.
Approach
The authors propose a Generative-Adversary-Trimodel (GAT) framework that pits a neural network (LLM) against an analytical model in an adversarial iteration to inject expert knowledge, verify constraints, and refine task sequences for autonomous disinfection.
Key results
- Achieves dual convergence between neural and analytical models, reducing logic, spatial, and temporal violations.
- Enhances planning efficiency for long-horizon disinfection sequences in dynamic settings.
- Enables zero-shot perception and execution of unknown object shapes and poses.
- Demonstrates higher success rates, shorter task times, and fewer rule violations compared to baseline methods.
Why it matters
Enables safer, more efficient autonomous disinfection robots for public health and clinical settings by bridging the gap between flexible AI reasoning and strict physical/temporal constraints.
Abstract
Robotic disinfection can relieve human operators from repetitive, labor-intensive tasks while reducing the risk of pathogen transmission in public spaces. Recent advances in learning-based methods further enhance these systems by enabling robust dynamic task planning and the interpretation of ambiguous instructions. However, disinfection task planning remains a four-dimensional (interaction, logic, spatial and tem- poral) problem that requires expert knowledge. The robust task planning for autonomous disinfection in dynamic environment remains challenging. This paper proposes a novel framework that integrating the Generative Adversarial Trimodel (GAT) method with embodied framework to solve the four-dimensional problem in the dynamic environment. The GAT method in- jects expert knowledge and iteratively refines neural network- generated plans against analytical model (AM), driving dual convergence and reducing logic, spatial, and temporal errors. By combining embodied framework and the GAT method into a GAT-enhanced embodied framework, the robot system autonomously perceives objects of unknown shape and pose, long-horizon task sequence plans, and executes disinfection operations. Experimental results demonstrate an improvement in success rate and reduce the average task time and rule vio- lation rates compared with non-GAT methods, demonstrating improved robustness and efficiency in dynamic environment.