CONVINCE SIT-L, a software pipeline for robot situation learning

SIT-L is a software pipeline for robot situation learning, part of the CONVINCE toolchain. It implements a pipeline for learning new situations. Those situations are unexpected failures that may happen at design time. This procedure is directly related to situation awareness, by tackling the detection, identification and mitigation of unknwon anomalies.

The module encapsulates situation awareness for detection and identification, added to it a learning procedure. In the near future WP3 active/simulate plan will be added to affine the learning.

Contents

• Deploy model (if needed) or use the local version
• Installations
  • Clone project
  • Build your project - Once
  • Activate virtual env - everytime you enter the project
  • Change environment variables (if you use a hosted model on a server)
• Format data (if needed)
• Learning (inference)
• Customize to your use case