Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Online road vehicle trajectory specification in presence of traffic lights with stochastic switching times

Volakakis Vasileios

Full record

---

Summary

The way someone is driving a road vehicle has an important impact on the fuel consumption, thus the term eco-driving was recently introduced to denote a driving style that reduces fuel consumption. This is correlated with many recent advances and developments that are taking place in vehicle communications and automated driving. One application of vehicle connectivity is to receive information about the next signal switching time, when a vehicle approaches a traffic light. Based on this information, appropriately developed systems, known as GLOSA (Green Light Optimal Speed Advisory), compute a fuel-efficient velocity profile for the vehicle to cross the traffic lights, e.g. without stopping, and provide drivers with speed advice.The main goal of this work is to generate optimal trajectories for vehicles crossing a signalized junction, with traffic signals operating in real-time (adaptive) mode. Specifically, the switching time of the traffic signal is decided, in real time, based on the prevailing traffic conditions and is therefore uncertain in advance. This extended (stochastic) GLOSA problem is addressed by using probabilistic traffic lights information and calculates a velocity profile for the vehicle based on the vehicle's initial state (position and speed) and a fixed final destination state.The problem is cast in the format of a stochastic optimal control problem, assuming availability of a time-window of possible signal switching times, along with the corresponding probability distribution, and is solved numerically using stochastic dynamic programming(SDP) techniques. As an ingredient of the stochastic solution, an appropriate deterministic optimal control problem is also formulated and solved analytically via Pontryagin’s Minimum Principle for the case of know switching times; the deterministic problem solution is used, as an initial trajectory for some extended SDP techniques that solve the problem in a significantly less amount of time compared to the standard SDP approach.The extended SDP techniques used in this work are the Discrete Differential Dynamic Programming (DDDP) method and the Differential Dynamic Programming (DDP) method. With these approaches, the workload and computational time are both significantly reduced, making the proposed approaches applicable in real time.

Available Files

Services