A causal-based approach to explain, predict and prevent failures in robotic tasks

Artikel i vetenskaplig tidskrift, 2023

Robots working in human environments need to adapt to unexpected changes to avoid failures. This is an open and complex challenge that requires robots to timely predict and identify the causes of failures in order to prevent them. In this paper, we present a causal-based method that will enable robots to predict when errors are likely to occur and prevent them from happening by executing a corrective action. Our proposed method is able to predict immediate failures and also failures that will occur in the future. The latter type of failure is very challenging, and we call them timely-shifted action failures (e.g., the current action was successful but will negatively affect the success of future actions). First, our method detects the cause–effect relationships between task executions and their consequences by learning a causal Bayesian network (BN). The obtained model is transferred from simulated data to real scenarios to demonstrate the robustness and generalization of the obtained models. Based on the causal BN, the robot can predict if and why the executed action will succeed or not in its current state. Then, we introduce a novel method that finds the closest success state through a contrastive Breadth-First-Search if the current action was predicted to fail. We evaluate our approach for the problem of stacking cubes in two cases; (a) single stacks (stacking one cube) and; (b) multiple stacks (stacking three cubes). In the single-stack case, our method was able to reduce the error rate by 97%. We also show that our approach can scale to capture various actions in one model, allowing us to measure the impact of an imprecise stack of the first cube on the stacking success of the third cube. For these complex situations, our model was able to prevent around 95% of the stacking errors. Thus, demonstrating that our method is able to explain, predict, and prevent execution failures, which even scales to complex scenarios that require an understanding of how the action history impacts future actions.