Skip to main content

Systems engineering, self-adaptation and robots with a deep understanding

Integrating advanced skills into robotic systems for reliable autonomous operation is a pressing challenge. Current robot software architectures, many based on the 3-layers paradigm, are not flexible to dynamically adapt to changes in the environment, the mission or the robot itself (e.g. internal faults).

In this talk, I will motivate the application of systems engineering in robotics to address this challenge.

I will present our results leveraging systems engineering knowledge for automatic reasoning to adapt the robot software architecture at runtime, with examples in mobile manipulators and underwater robots.

To conclude, I will briefly present our upcoming projects, CoreSense and METATOOL, where we will develop hybrid cognitive architectures with deep understanding and self-awareness capabilities, hopefully resulting in robots that are more flexible, reliable and explainable, and that are capable of tool invention, respectively.

Lecture Video