The main objective of Dreams4Cars is to set up an offline simulation mechanism in which robots, by recombining aspects of real-world experience, can produce a simulated world, with which they can collectively interact to safely develop and improve their Perception-Action systems, in particular focusing on the analysis of rare events.
The application domain of Dreams4Cars is automated driving which also poses the issue of developing systems capable of dealing with arbitrary and open-ended circumstances. The main objectives are:
Implement an artificial driver agent with a Perception-Action (PA) system capable of hosting extendable/reconfigurable behaviours (using layered control architecture with action selection). Implement forward emulators producing an emulated world (using combined machine learning and multi-body system dynamics).
Implement a technology that discovers and optimises behaviours by means of simulations carried out offline, and which communicates adaptation for online use to all the artificial drivers.
Demonstrate the effectiveness of this technology by evolving one driving agent with cycles of activity and simulation (‘dreams’) on research-grade vehicles.
Port the same agent to a real production vehicle and demonstrate the achievement of TRL 6, as well as an increase in the level of abilities compared to the baseline of a state of the art human engineered system.