Use of VAE to identify anomalous data in robots

Use of VAE to identify anomalous data in robots

Use of VAE to identify anomalous data in robots 150 150 UKAEA Opendata
UKAEA-RACE-PR(22)03

Use of VAE to identify anomalous data in robots

For robotic systems involved in challenging environments it is crucial to be able to 2 identify faults as early as possible. In challenging environments it is not always possible to explore 3 all of the fault space, thus anomalous data can act as a broader surrogate, where a anomaly 4 may represent a fault or a predecessor to a fault. This paper proposes a method for identifying 5 anomalous data from a robot, whilst using minimal nominal data for training. A Monte-Carlo 6 ensemble sampled Variational Autoencoder is utilised to determine nominal and anomalous 7 data through reconstructing live data. This has been tested on simulated anomalies on real data, 8 demonstrating the technique being capable of reliable identifying anomaly, with no pre-knowledge 9 of the system. With the proposed system getting an F1-score of 0.85 in testing.

Collection:
Journals
Journal:
Robotics
Publisher:
MDPI (Multidisciplinary Digital Publishing Institute)
Published date:
21/07/2021