Project Summary
The proposed innovation adapts a novel energy optimisation model to provide eco-driving speed recommendations to drivers in real-time.
Project Achievements
At UCL we used a T-TRIG grant to apply machine learning to the original eco-driving problem. We assessed the effectiveness of the strategy in microscopic traffic simulation on an urban scenario in the UK. This allowed us to determine possible energy savings in autonomous vehicles which always followed the recommendations, and in a driving simulator where real life drivers were provided live recommendations on a screen.
Conclusions
An effective eco-driving strategy was developed which can provide speed recommendations in 0.002s. Analysis of this in a simulated environment demonstrated considerable energy savings, with 15% energy savings in high-density traffic and 9% in low-density traffic. Additionally, no severe impact on travel time or energy usage of other road users was observed. Savings of this magnitude can be achieved if used in autonomous vehicles to select the driving speed.
In order to assess the viability of the strategy as a driver guidance system, a simulator study will be performed to assess the users’ acceptance and optimize the way that eco-driving information can be conveyed. Using the T-TRIG grant we developed software for UCL’s driving simulator that creates a graphical environment which represents a real-word road layout and can generate simulations without requiring any additional programming. This gives users an easy and quick way to develop test cases that can be readily used for experiments.
