The main goal of the Green Co-Driver project is to improve the energy efficiency of Connected and Autonomous Road Vehicles in ways that are highly acceptable to humans. This is achieved by integrating a green driving function and an interactive self-driving agent into a novel Green Interactive Self-Driving System (Green Co-Driver), which promotes energy efficiency, respects drivers’ preferences, and ensures safety.

The Green Co-Driver employs a specialized sensorimotor architecture (a layered cognitive control architecture) with energy-aware motor primitives (i.e., new layers of the sensorimotor stack that assess the energy footprint of potential actions). It continuously generates a large pool of safe maneuvers. From this pool of alternatives, the agent selects the best trade-off between travel time and energy efficiency. The action-selection process may be influenced by human input, such as actions involving pedals and steering wheel. In this way, the human driver can influence the balance, gradually prioritizing speed while remaining on the Pareto frontier of minimum cost. The interaction between the Green Co-Driver and the human driver will be assessed using both a static simulator (UNITN) and a dynamic one (DriSMi).