TIMID: Time-Dependent Mistake Detection in Videos of Robot Executions
This addresses the challenge of identifying temporal errors in robotic systems, which is crucial for improving reliability in real-world applications, though it is an incremental advance building on VAD frameworks.
The paper tackles the problem of detecting complex time-dependent mistakes in robot task executions from video, which existing Video Anomaly Detection methods struggle with, and introduces TIMID, an architecture that achieves frame-level prediction of such mistakes using weak supervision and a new simulation dataset.
As robotic systems execute increasingly difficult task sequences, so does the number of ways in which they can fail. Video Anomaly Detection (VAD) frameworks typically focus on singular, low-level kinematic or action failures, struggling to identify more complex temporal or spatial task violations, because they do not necessarily manifest as low-level execution errors. To address this problem, the main contribution of this paper is a new VAD-inspired architecture, TIMID, which is able to detect robot time-dependent mistakes when executing high-level tasks. Our architecture receives as inputs a video and prompts of the task and the potential mistake, and returns a frame-level prediction in the video of whether the mistake is present or not. By adopting a VAD formulation, the model can be trained with weak supervision, requiring only a single label per video. Additionally, to alleviate the problem of data scarcity of incorrect executions, we introduce a multi-robot simulation dataset with controlled temporal errors and real executions for zero-shot sim-to-real evaluation. Our experiments demonstrate that out-of-the-box VLMs lack the explicit temporal reasoning required for this task, whereas our framework successfully detects different types of temporal errors. Project: https://ropertunizar.github.io/TIMID/