Hybrid-state driver/vehicle modelling, estimation and prediction

Predicting driver maneuvers by learning holistic features

“…Vehicle dynamics are used in [7] to identify driver intent at intersections and recognize actions. Vehicle trajectories were used in [8] to model driver intent at intersections using a Probabilistic Finite State Machine. Although methods commonly incorporate ego-vehicle dynamics, others may take on a purely vision approach.…”

HMM and HSS Based Social Behavior of Intelligent Vehicles for Freeway Entrance Ramp

“…What's more, the changes of continuous states also in turn affect the discrete states. Therefore, the system architecture of intelligent vehicles is a kind of Hybrid state system [13][14], which combines continuous state system and discrete state system together, as shown in Figure 6. As shown in Figure 6, the working process of the whole system architecture is as follows: a) The high level control model carries out the control strategy based on social vehicles' intention  , intelligent vehicles' information s , and the control strategy X by rule-based FSM; b) State handling in the information process converts the control strategy into driving command S (route and speed)，and pass it on to the low level controller; c) The low level controller obtains the new continuous driving states based on driving command S and the position & orientation information y so as to realize the control of intelligent vehicles.…”

Exploiting map information for driver intention estimation at road intersections

“…In [10] the basic elements constituting manoeuvres are recognised using fuzzy rules combined with a Bayesian filter, and a Probabilistic Final State Machine (PFSM) is used to represent the possible sequences of basic elements of several driving manoeuvres. The approach proposed in [11] is to model driver intention at an intersection using a PFSM where the transition probabilities are set dynamically using the output of a continuous vehicle state tracker. These works do not take into account contextual information; they infer driver intention from vehicle state only.…”