The project was focused on incremental autonomous learning, experience transfer, and robust avoidance strategy, with the virtualization of a real‐world robot to learn reaching a target while avoiding obstacles in a simulation environment.

Virtualization of a real‐world robot: 3D mesh > movable parts > joints and direct kinematic chain



Learning to reach a target while avoiding obstacles in a simulation environment with incremental autonomous learning, with experience transfer and robust avoidance strategy:

 

Related publications

• Bianca Sangiovanni, Gian Paolo Incremona, Marco Piastra, Antonella Ferrara (2020). Self-Configuring Robot Path Planning With Obstacle Avoidance via Deep Reinforcement Learning, in IEEE Control Systems Letters, pp.(99):1-1, DOI: 10.1109/LCSYS.2020.3002852
• Bianca Sangiovanni, Gian Paolo Incremona, Antonella Ferrara, Marco Piastra (2019). Deep Reinforcement Learning Based Self-Configuring Integral Sliding Mode Control Scheme for Robot Manipulators, 57^th IEEE Conference on Decision and ControlAt, Miami Beach, FL (USA), DOI:10.1109/CDC.2018.8619843.
• Bianca Sangiovanni, Angelo Rendiniello, Gian Paolo Incremona, Antonella Ferrara, Marco Piastra (2018). Deep Reinforcement Learning for Collision Avoidance of Robotic Manipulators, 17^th European Control Conference, Limassol, Cyprus, DOI:10.23919/ECC.2018.8550363.

See also: I robot Epson entrano all’Università di Pavia per imparare a interagire con l’uomo (Video), from UNIPV.news (2020).