2.3 Two-Dimensional Microscopic Models
2.3.6 Summary
Several modifications in the existing vehicle-following models have been made by different researchers by incorporating vehicle heterogeneity and lateral separation effects. More studies have been conducted by modifying GHR model and optimal velocity model. Researchers have incorporated vehicle heterogeneity (trucks, cars) and identification of governing leader using the GHR model framework. In a similar manner, some of the researchers have considered the effects of lateral separation into optimal velocity model. This model is only suitable to capture the dynamics of the evolution of traffic congestion, stop and go waves, etc. but it cannot be used to describe uninterrupted traffic. Although CA model can provide a better representation of mixed traffic and easier to implement, the main disadvantage of the model is a loss of detail because of the discretisation. A major contribution in describing the mixed traffic stream was made by Gunay (2007), by considering lateral discomfort in the Gipps car-following model.
Though the variability in road geometry as well as the effect or road edge in considered by Chakroborty et al. (2004), the variability in edge conditions are not included. The available CA models do not mimic the dynamic behaviour of the vehicles on the real world traffic stream.
Hence, it results in misestimating the microscopic behaviours of drivers observed in the mixed traffic stream.
In the case of Indian traffic conditions, the roads are having different types of geometric features with various types of edge conditions. Most of these models did not represent the driving behaviour of the mixed traffic stream. This is due to the absence of proper
comprehension of drivers’ behaviour in mixed no-lane based traffic streams. A comprehensive and complete model still need to be developed. The performance of microscopic model is mainly depends on accuracy of the data obtained from the real world traffic stream (field study).
Due to weak lane discipline, vehicles manoeuvring in mixed traffic exhibit complex driving patterns of maintaining shorter headways, tailgating, swerving and filtering, etc. This complex driving behaviour can be modelled by the data collected from the real traffic stream. It is always a challenge for the traffic modellers to use an adequate number of parameters for describing the driving behaviour precisely with simplifying the implementation and calibration process. Hence a brief study on the real traffic data of heterogeneous no-lane based traffic is important.
Table 2.2 Summary on the modification made in basic CF models for mixed traffic conditions
Basic CF model
Vehicle
types Type of study Parameters added Data
modelled Calibration Simulation
platform Final model representation Reference
GHR Model C, T
Single lane car-following model considering
acceleration and deceleration responses
Different leader-follower pairs were considered:
C-C, C-T, T-C
NGSIM 45- min data
Non-linear
regression -
Parameters for space headway, speed, and relative speed were estimated.
(Siuhi &
Kaseko, 2016)
HMV, LMV, 2W Investigating governing leader from
Direct front vehicle
Left front vehicle
Right front vehicle
Choice of governing leader using discrete choice model (space headway, lateral overlap, subject vehicle type, relative speed)
9 vehicle-pair combinations were considered to determine the deceleration/acceleration responses
90-min trajectory data
from Dhaka
Broyden–Fletcher–
Goldfarb–Shanno (BFGS) an algorithm using
Oxmetrics 6
-
Deceleration functions for different vehicle pairs were estimated;
Acceleration function for all vehicles
(Choudhur y & Islam,
2016)
Safe distance Model Cars
Staggered car-following One right front leader
LS, FC
Maximum Escape speed (speed required for passing through the escape corridor)
Veering distance
Multilane highway in
Istanbul, Turkey
Assumptions of Gipp’s model
C++ Space-time trajectories, Gunay (2007)
Speed, Flow, Density diagrams (Gunay, 2009)
(Gunay, 2007) (Gunay,
2009)
Staggered car-following
Right front vehicle
Left front vehicle
FC is proportional to escape speed
FC is inversely related to following time and minimum following safe distance.
CS separation with the leader
2-lane Matlab Space-time diagrams for different
FC, CS.
(Xu, 2015)
C, 3W, T, B
Staggered car-following considering swerving manoeuvre (when duration of swerving is less than the reaction time)
Lateral clearances of LV and FV
Maximum lateral speed of FV
Near Cadbury Junction, Thane at evening peak
hour
AIMSUN simulator
15-minutes traffic counts were compared
(Lenorzer et al., 2015)
Different leader-follower
vehicle types in Gipp’s model Reaction time, deceleration, maximum acceleration, desired speed of FV
Assumed deceleration, effective size of LV
Sensitivity-gap maintaining behaviour of different vehicle pairs
Single lane, NGSIM data and Mumbai
data
Genetic Algorithm C Simulated speed-flow diagrams for different vehicle pairs
(Ravishank ar &
Mathew, 2011)
Basic CF model
Vehicle
types Type of study Parameters added Data
modelled Calibration Simulation
platform Final model representation Reference
Optimal velocity Model Cars
Staggered car-following Two leaders :
Direct front vehicle
Left front
LS effect between the leader and follower.
Visual angles
Single lane Trial and error Numerical simulation
Space-time trajectories Headway and velocity variations for different lateral separation parameter
(Jin et al., 2010) (Zheng et al., 2012) Staggered car-following
One Right front leader
Time-to-collision variable using visual angle and visual gap angle.
Headway and speed evolution
with time (Jin et al.,
2012) Staggered car-following
Direct front
Left front
Right front
Two sided LS effects
LSn for the right front and left front vehicles are equal.
𝑝𝑛 = 0, 0.5, 1 are taken
Space headway of cars at
different time steps (Li et al., 2015) Staggered car-following
Two leaders :
Direct front vehicle Left front vehicle
LS between the subject and the leader
Escape corridor
NGSIM 25-min
data
Genetic Algorithm
Comparison of Gap, velocity variations with time for proposed model, OVM and IDM
(He et al., 2015)
Note: C: Cars, 2W: Motorized two-wheelers, 3W: Motorized three-wheelers, T: Trucks, B: Buses; HMV: Heavy Motor Vehicles, LMV: Light Motor Vehicles; LS: Lateral Separation, FC: Frictional Clearance, CS: Centreline Separation; NLBCF: Non-Lane Based Car-Following model, FVD: Full Velocity Difference model, OVM: Optimal Velocity Model; IDM: Intelligent Driver Model; NGSIM: Next Generation Simulation (US Highway 101 dataset); AIMSUN: Transportation Simulation Software