Intelligent Execution through Plan Analysis

TL;DR

Robots typically plan under fixed world models, but execution frequently reveals opportunities to improve plans rather than requiring replanning. The authors introduce dynamic opportunities derived from causal links between actions, compute them at planning time, and deploy an intelligent monitoring loop that detects these opportunities during execution and repairs the plan accordingly. The approach, clo, focuses perception on relevant changes and repairs only the affected portions of the plan, outperforming standard replanning across Rooms, Dialog, and Cooking domains. This yields substantial reductions in planning overhead and improves execution efficiency in open-world robotic tasks.

Abstract

Intelligent robots need to generate and execute plans. In order to deal with the complexity of real environments, planning makes some assumptions about the world. When executing plans, the assumptions are usually not met. Most works have focused on the negative impact of this fact and the use of replanning after execution failures. Instead, we focus on the positive impact, or opportunities to find better plans. When planning, the proposed technique finds and stores those opportunities. Later, during execution, the monitoring system can use them to focus perception and repair the plan, instead of replanning from scratch. Experiments in several paradigmatic robotic tasks show how the approach outperforms standard replanning strategies.

Figures (2)

• Figure 1: Example of the Rooms domain with one robot, two objects and three rooms. The goal is to grab all objects.
• Figure 2: Total time in the (a) Rooms, (b) Dialog and (c) Cooking domains taken by clo (replanning with opportunities) and replan.

Theorems & Definitions (1)

• Definition 1: Causal link