Follow Me: Hierarchical Parallel Execution Synchronization in Behavior Trees

Abstract

The Behavior Tree (BT) has gained considerable popularity in many areas such as robotics and artificial intelligence, due to its advantages of modularity, reactivity, reusability, readability, etc. In the classic BT, tasks under the Parallel node are assumed to be independent of each other and cannot be executed synchronously. To address this problem, we introduce hierarchy into parallel tasks by dividing parallel tasks into leader and follower tasks. In addition, we define the state value for each task node to achieve efficient synchronization between tasks. We also propose two new parallel operators to solve the synchronous execution of parallel tasks in BTs. We integrate our approach into a popular BT framework and evaluate it by multiple experiments. Results show that our approach improves the parallel execution in traditional BT models and has great practical significance.