18 Nov 2022
Researchers from the University of Waterloo, DarwinAI, and Moog Inc. developed SolderNet, an explainable AI system for automated solder joint inspection. This system accurately classifies solder joints as defective or non-defective while providing visual explanations for its decisions, aiming to enhance transparency and trustworthiness in electronics manufacturing quality control. The Attend-NeXt Large model achieved 91.1% accuracy with a 5.0% overkill rate and low escape rates (0.6-2.3%).
14 Dec 2024
TERA is a comprehensive simulation environment designed for developing and evaluating end-to-end autonomous excavation systems. It integrates high-fidelity soil interaction, realistic robot dynamics, and extensive sensor simulation, demonstrating strong sim-to-real transfer capabilities with a Root Mean Square Error of 1.376m for navigation trajectories.
There are no more papers matching your filters at the moment.
