Bootstrapping Self-Supervised Learning of Binary Classification Using Error Bounds: A Case Study on a Robotic Insertion Task
2026-07-31 • Robotics
Robotics
AI summaryⓘ
The authors focus on making flexible manufacturing faster and safer by reducing setup time and controlling errors. They developed a system with two parts: a quick model that makes predictions and a slower, more accurate check that verifies those predictions when needed. By measuring how confident the quick model is, the system decides when to ask for a detailed check, helping balance speed and accuracy. Their method was tested on a robot insertion task using force data, and over time, the system improved and needed fewer expensive checks while keeping errors low. This shows how confidence measurements can help robots learn and work more reliably.
Flexible manufacturingModel predictionConfidence levelVerificationRobotic insertionForce dataUMAP dimensionality reductionWilson-ScoreError controlSelf-improving models
Authors
Zebin Duan, Norbert Krüger, Juan Heredia, Thorbjørn Mosekjær Iversen, Frederik Hagelskjær
Abstract
Flexible manufacturing requires rapid deployment of solutions and minimal setup time to remain competitive. An essential attribute is the ability to control error levels, as failures can range from minor performance degradation to severe equipment damage. However, conventional deployment often involves extensive setup, data collection, model training or parameter tuning, and system testing, resulting in significant delays that hinder commercial feasibility. We propose a data engine which gathers data and improves its performance while executing the task. The data engine consists of two classifiers, a fast model prediction and expensive verification. First, a model prediction is performed and based on the confidence level of the prediction, the expensive verification can be used. By adjusting the confidence level, users can control the level of tolerable error. Our method is implemented on a real-world robotic insertion task, which uses force data for the model prediction. The system applies UMAP dimensionality reduction and uses Wilson-Score to compute the confidence bounds of the prediction. Results demonstrate the ability to learn and reduce the need for expensive verifications over time, while staying within the set error-rate. The results highlight the potential of confidence bounds in self-improving models to enhance reliability in robotic classification task.