Control and decision method for overtaking problem in autonomous vehicles

Hegedűs, Tamás and Németh, Balázs and Bede, Zsuzsanna (2018) Control and decision method for overtaking problem in autonomous vehicles. In: 16th Mini Conference on Vehicle System Dynamics, Identification and Anomalies.

Text Control and decision method for overtaking problem in autonomous vehicles.pdf Restricted to Repository staff only Download (772kB)

Abstract

In this paper we deal with an overtaking and lane change strategy for autonomous vehicles. The algorithm is divided into two main layers. The upper layer is responsible for the decision making (e.g lane changing). Before decision-making we predict the motion of other participants and divide the road into discrete sets in each interval, then we compute the probability of occupancy. Considering the state of the motion of ego vehicle we can compute the probability of collision. After assigning a graph points to discrete elements, the probability of collision is ordered to the graph edges, then we determine the route with the smallest cost. The decision-making algorithm is not able to control the vehicle, so it is necessary to design a traceable trajectory for the vehicle. Since we have a series of points to be reached for the future, this can be used as a reference vector for the trajectory design algorithm. One Model Predicitve Controller (MPC) is responsible for the design of the trajectory so we can obtain limitations to several states to ensure comfort and security. Extending the algorithm, we can compute the longitudinal movement as well. In this paper the decision-making and the trajectory planning algorithm is presented, and the whole model validated in CarMaker.

Actions (login required)

Edit Item