vehicle actuators input to the vehicle controller. The
vehicle controller itself consists of several feedback
and feedforward controllers to guarantee that the
vehicle follows the planned trajectory. Another
important part of the decision-making module, is the
cooperative cruise controller which calculates a
velocity for the vehicle based on the information
received about the preceding vehicle via V2V
communication. Driving with that velocity results in
driving with shorter headway to the preceding vehicle.
The majority of the research on the idea of
platooning has been conducted in a highway-based
situation. However, recently the research work has
turned towards platooning in urban areas, where
platooning is mostly linked to efficient intersection
passing rather than reducing air drag. Although
requiring a high amount of flexibility, the idea of
urban platooning has already been tested in public
traffic (Schindler, et al., 2020) (Dariani & Schindler,
2019). However, it is still far from being normalized
or standardized. The communication network needed
for cooperation in this paper is only based on the
preceding vehicle and no other information such as
leader information is required. That makes the
cooperation very dynamic especially in urban areas in
which the string of the vehicles mostly does not have
a common destination and the vehicles drive together
only for few intersections. In this case forming and
resolving a platoon is very dynamic and adaptive to
urban area scenarios.
Figure 1: String of vehicles driving with CACC.
The main focus on this paper is on the trajectory
planner and the decision making. Although the
decision-making modules focuses on many aspects
such as behaviour and intention prediction of other
participants as well as analysing road geometry
(Dariani & Schindler, 2019), in this paper only
platooning related functionalities of the decision-
making module are discussed.
The outline of the paper is as follows, chapter 2
describes the vehicle automation and briefly explains
the trajectory planner and decision-making module.
In Chapter 3 the trajectory planner is explained.
Chapter 4 is about the decision-making module with
the focus on the platoon management module and the
cruise controller. In Chapter 5 the functionality of the
designed algorithms has been proven in simulations
and tests in public traffic in a complex urban area, and
finally Chapter 6 is conclusion.
2 VEHICLE AUTOMATION
The Automated Driving Open Research (ADORe)
developed by the Institute of Transportation Systems
of the German Aerospace Center (DLR), also
available open source (Hess, et al., 2017), is a
modular software library and toolkit for decision
making, planning, control and simulation of
automated vehicles has been used for this work, see
Figure 2. As the same software is used in simulation
and in research vehicles, the simulation experiments
are very close to reality. Although many modules
remain unchanged in this work such as Navigation,
Controller, Data Model, etc., several modules have
been completely changed or modified explicitly for
this research work, such as Decision-Making,
especially the platoon management module,
Trajectory Planning and cruise controller.
Figure 2: ADORe modular architecture.
For Trajectory planning an optimal control
approach is used which makes the planned trajectory
the solution of a nonlinear optimization problem. One
powerful method to solve a sequence of nonlinear
Optimal Control Problems (OCP) is Sequential
Quadratic Programming (SQP). The Newton method
or quasi-Newton method finds a point where the
gradient of the objective function of the OCP
vanishes. The Newton or quasi-Newton method
requires a starting point or an initial solution and the
quality of the initial solution has high impact on the
convergence rate of the optimization problem and
consequently on the calculation time. Therefore, an
initial solution is calculated based on the shortest path
connecting current vehicle position to destination,
which is already available via “Navigation” module.
A “Decision-Making” module is designed on top of
the trajectory planner to define the strategical and
tactical tasks for the planner, i.e. the long- and short-
term tasks. Mainly due to the complexity of the non-
linear optimization problem, the planning horizon, ,
has its real-time limitation and cannot merge to
infinite. But the decision-making horizon can be
extended to the vehicle perception sensors vision
range or even to the communication range, which
permits the trajectory planner to take required actions