Monitoring Safety in Robot Manipulation

Our recent research focuses on the development of a deep multimodal sensor fusion framework (FINO-Net) for failure/anomaly detection and prediction in manipulation tasks by humanoid robots.

We have expertise in multimodal sensor fusion methods for processing RGB, depth and audio readings and deep learning methods to effectively detect robot manipulation failures/anomalies.

Project page: https://air.cs.itu.edu.tr/projects/finonet.html