تطوير نموذج تمثيل هرمي موحد للتخطيط والعمل في الروبوتات المستقلة

Developing a unified Hierarchical Representation Model for Planning and Acting in Autonomous Robots

Hierarchical frameworks have traditionally been used by humans to organize their thoughts on complex problems. One application of artificial intelligence involves the creation of intelligent agents that break down challenging problems into layers of abstraction, simplifying the problem-solving process. The Hierarchical Task Network (HTN) structure enables tasks to be effectively performed.
In the context of autonomous robots operating in closed and deterministic environments, such as domestic environments, task planning is crucial for achieving a high level of accuracy. To address this, the HTN planner and descriptive action models were employed to determine the next state in the state transition system during the planning process for the generated plan. A formal representation for operational action models was introduced, and their effective utilization in conjunction with the Refinement Acting Engine (RAE) and the Dijkstra algorithm to determine optimal task approaches was explained for the execution of this plan.
The integration planning and acting algorithm was designed to operate hierarchically, utilizing a transit tree and backtracking to handle the error for re-planning and execution in the event of execution errors. Careful attention was given to method implementation to avoid system failure or suboptimal performance.
The methodology used in the text involves addressing the task planning challenges faced by autonomous robots operating in closed and deterministic environments, specifically domestic environments. To achieve a high level of accuracy in task execution, the Hierarchical Task Network (HTN) planner and descriptive action models were employed with representation for operational action models the Refinement Acting Engine (RAE), and the Dijkstra algorithm to determine optimal task approaches, The hierarchical structure and error handling mechanisms contribute to achieving accurate and efficient task completion while maintaining system reliability depending on The integration planning and acting algorithm was designed to operate hierarchically, utilizing a transit tree and backtracking algorithm.
Through experiments conducted using the Python language (Spyder), it was demonstrated that the algorithm developed outperforms the commonly used planning and acting approach, providing an optimal path within the system. The results showed a precision value of approximately 94.9%, a recall value of approximately 94.9%, and an F1-score of approximately 94.79%. These findings validate the effectiveness of the approach in improving the planning and acting process.