# MSc projects

## MSc thesis topics

- Traffic flow theory for drones
- Traffic control for drones
- Macroscopic FD of warehouses
- Control of traffic in large regional road networks
- Lane change control with parrallel roadways (with Autonomous vehicles)
- Autonomous vehicles and stability for multi-leader systems
- Autonomous vehicles: Platoon forming
- Autononous vehicles: lane usages
- Effects of keep-your-lane (with autonomous vehicles)

## Finished projects

- Speed limits and their effect on freeway capacity
- Ramp metering: a microscopic control approach
- Representing the Car-Following Behaviour of Adaptive Cruise Control (ACC) Systems Using Parametric Car-Following Models
- Modelling traffic in the Randstad using a dynamic zone model based on the Network Fundamental Diagram
- New Intersection Control for Conventional and Automated Vehicles without Traffic Lights
- Flow Capacity of Bottlenecks in a Cycle Storage
- Using relative flow data to expose the fundamental diagram

## MSc thesis topics

### Traffic flow theory for drones

A future increase of flying objects, including drones, could potentially change the way the airspace is being used. Instead of planning individual flights with sufficient space, there could be streams of drones, possibly in various directions in 3 dimensions. The space need to be optimized to enable large flows, and short routes. Within this thesis you will test how air traffic can be described.

The objective of this thesis is how traffic flow theory, which is originally developed for 1 dimensional (road bound) traffic and which is now being modified for 2 dimensional (pedestrians) traffic, can be applied to 3 dimensional traffic. Do fundamental diagrams and flow equations hold? You can test various systems of regulations: planned flights or autonomous agents.

**Supervision**: dr. Victor Knoop and dr. Joost Ellerbroek (aerospace engineering)

### Traffic control for drones

Currently, air traffic is highly regulated. For each flight, a flight plan is made and manually accepted. A future increase of flying objects, including drones, could potentially change the way the airspace is being used. The way conflicts are handled should potentially be changed, for instance to a lower level. This is comparable to car traffic, where not all trips are planned, but control is made at local intersections by traffic lights or by priority rules.

In this thesis, the main question is (a) which road traffic regulations can be transformed to drone traffic and (b) what the efficiency of the traffic under these regulations is. One idea behind transforming is that differences in timing (for instance in a traffic light) can be transformed to different altitude levels for drones. Exploiting these symmetries can be useful.

This research question is approached combining disciplines in civil engineering and aerospace engineering, and is suited for a student wanting to straddle on interdisciplinary boundaries.

**Supervision**: dr. Victor Knoop and dr. Joost Ellerbroek (aerospace engineering)

### Macroscopic FD of warehouses

In logistics, optima need to be found to arange and move freight. Storage and handling capacity in warehouses (or harbors) might be a constraint, limiting movement and increasing turnover times. In the past decade, the network fundamental diagram has been discovered and studied in traffic flow theory. The main idea is that one does not need to go into detail of individual vehicles or roads, but there are relationships between number of vehicles and speeds at the aggregate level. Also, if a critical number of vehicles in the network is exceeded, that can decrease the outflow.

**Objectives & Assignment**

In your assignment, the main question is whether a relationship between number of objects and turnover time holds for warehouses. Also the aim is to find this relationship, including the characterizing variables (e.g., size of the warehouse, number of handling agents, ....) The assignment will consist of modelling of operations at a warehouse, possibly combined with analysing data. Placement at a company is possible.

**Supervision**: dr. Victor Knoop and prof. dr. L. Tavasszy

### Control of traffic in large regional road networks

Different control strategies for roads are available. If complexity increases, the interactions become too complex to describe the consequences of a control action in detail. In an earlier thesis work, a multi-zone model is developed to describe the traffic in a large area like the Randstad (the multi-centered western part of the Netherlands, 8 million inhabitants). For each of the centers, it uses bins, for which an aggregated traffic description is used.

On this level, can we develop and test control measures? And what is their effect? Some are easy to be modelled, e.g. reducing traffic demand. Others are harder, like for instance peak hour lanes or adapting traffic light settings. How would these be implemented on this model scale, and what is their effect?

**Supervision**: dr. Victor Knoop

### Lane change control with parrallel roadways (with Autonomous vehicles)

The distribution of traffic over lanes on a freeway is not equal: some lanes carry more traffic than others. Recent research explores the options to control traffic over lanes. Basically, steering traffic away from dense lanes and/or away from merging zones can help increase the capacity of the road and hence reduce delays.The research did not include trucks, which form a special class of cars which have a different speed and cannot change lanes easily.

In this thesis, special attention will be given to trucks. Also, a wider variaty of road structures is used. Especially a structure where a road near an city splits into two roadways is considered. What type of advice is most suitable here?

A variation to the thesis topic could include the use of (some) fully autonomous vehicles, for which not only advice can be given, but they can be fully controlled.

**Supervision**: dr. Victor Knoop

### Autonomous vehicles and stability for multi-leader systems

Current vehicles are to a large extent autonomous. Whereas the supervision still lies with the driver, they can drive themselves within a lane in the motorway. That means that they themselves determine the acceleration as response to the vehicle in front. Recent studies show that these ACC systems are not so-called string-stable, i.e. if the first vehicle brakes a bit, the second brakes a bit more, the third even more etc. If there are groups (platoons) of approximately 7 vehicles following each other using ACC this leads to dangerous situations. In the future, with a higher penetration of ACC systems, this needs to change.

One option to change this is to include not only the information (speed) of the leader, but also of the leader of the leader (second leader). Technically, this is possible. Within this thesis project, you try to include this information in the response of the ACC. How close can vehicles drive without creating unstable platoons?

**Supervision**: dr. Victor Knoop and Meng Wang

### Autonomous vehicles: Platoon forming

Vehicle automation is increasing and vehicles are doing the driving more and more themselves. Besides, communication between the vehicles is also an emerging. This enables many new strategies for driving.

Within one lane, specific arrangements of vehicles following each other can be beneficial (i.e., vehicle X should be leading, then vehicle Z, then vehicle A). This might have for instance effects on the stability of traffic streams. The stability of a traffic stream is also influenced by lane changes. They can work as disturbance. At the other hand, they can also provide necessary relief for a disturbance.

Within this thesis work, you;ll explore this field. If platoons of autonomous vehicles (and perhaps also human-driven vehicles…?) should be made, how can that be done in the best way? Which rules should be followed. This includes designing rules and testing them in simulation.

**Supervision**: dr. Victor Knoop (possibly combined with Meng Wang)

### Autononous vehicles: lane usages

Current vehicles are to a large extent autonomous. Whereas the supervision still lies with the driver, they can drive themselves within a lane in the motorway. That means that they themselves determine the acceleration as response to the vehicle in front. “Lazy” drivers might not interfere with the system and stay in their own lane.

Within this thesis you will explore experimental data for the lane usages and lane changes of drivers of vehicles equipped with ACC and lane keep assistance (lane centering systems). Moreover, and a large extent of the thesis work: you will check the effects of those (changed) behaviors for the traffic stream. This is most likely done by microscopic simulation.

**Supervision**: dr. Victor Knoop

### Effects of keep-your-lane (with autonomous vehicles)

European regulations require to keep right unless overtaking. In some states in the USA, this is not the case, and traffic can keep it’s desired lane. This might save lane changes, and therefore increase capacity and increase safety. At the other hand, empirical capacity values suggest that the capacity in the Netherlands is higher than in the US.

In this thesis, the system of keep-your-lane is simulated and compared to a “keep right unless overtaking” system. Important is to analyse underlying behavior from real-world data such that the only change is the lane discipline. Using the simulation, conclusions on effectiveness and safety are to be draw. In this thesis, also policy effects can be analysed (TPM background, suited for TIL).

**Supervision**: dr. Victor Knoop

## Finished projects

### Speed limits and their effect on freeway capacity

**By: ****William van Lindonk **

In this thesis an investigation is performed into the effect of different speed limits on freeway capacity. From literature, much is known about the variety of factors that affect capacity, but the exact effect of the speed limit on capacity is not yet clear. In recent years, several speed limits changes have taken place at multiple two-lane freeway bottlenecks throughout The Netherlands, which makes it possible to compare effects of different speed limits at the same location. To evaluate the effect of the speed limit on capacity, the Product Limit Method has been applied to identify breakdown flows and generate capacity distributions, which could subsequently be compared for different limits. In the comparison of capacity distributions under different speed limits, it was found that significant changes in capacity had occurred, but that no uniform direction of the effect could be found. Subsequently, to control for location specific factors and other variables, Fixed Effects regression has been used to determine the effect of the speed limit on the breakdown flow. It was found that the breakdown flow under the 120 km/h limit was significantly higher than under the 130 km/h limit (in the range of 60 to 190 vehicles per hour) and that the breakdown flow under the 100 km/h limit was, in some cases, also higher than under the 130 km/h limit. In addition to this, it was found that a significant positive relation exists between the height of the speed limit and the fraction of flow in the passing lane. Moreover, it was shown that the relation between the fraction of flow in the passing lane and the level of breakdown flow was best represented by a quadratic relation, which could indicate that an “optimal” distribution of flows may exist. Given the results of this thesis, it is posed that a change in the speed limit is likely to affect capacity primarily through altering the lane flow distribution and that it will depend on the layout of a freeway location what the optimal lane flow distribution is and which speed limit leads to this optimal lane flow distribution.

### Ramp metering: a microscopic control approach

**By: ****Stefan Klomp **

Freeway on-ramp areas are susceptible to traffic congestion during peak hours. As a result, many stop-and-go waves on freeways are caused at these sites. To delay and/or prevent the onset of congestion, ramp metering can be applied. This means the inflow to the freeway is limited by traffic lights at the ramps. Ramp metering controls the inflow from the on-ramp onto the main line so that the total flow can be kept just below capacity. Current ramp metering algorithms however, are based on macroscopic traffic characteristics like the mean flow in one minute. These type of algorithms do not avoid inefficient merging manouvers entirely. In this thesis, a microscopic ramp metering approach is proposed, based on gap detection on the right lane of the main line.

First, the acceleration characteristics of accelerating vehicles are studied. For this, video is captured at an existing on ramp. Then, this behavior is implemented in a microscopic simulation and various control algorithms are tested. This way, we found that microscopic ramp metering approach yields more travel time savings for the entire system for a number of scenarios. For the base case, it was found that a small gain in travel time savings is realized for vehicles on the main line when the microscopic control approach is compared to the reference algorithm. Additionally, travel time savings are also obtained for vehicles coming from the on-ramp. Especially during periods of very high main line demand, the microscopic control approach is able to achieve additional travel time benefits. There is still room for improvement of this microscopic algorithm, enabling even more travel time savings. Combining the traditional algorithm with the microscopic approach should have the potential to increase the effectiveness of the microscopic control approach.

### Representing the Car-Following Behaviour of Adaptive Cruise Control (ACC) Systems Using Parametric Car-Following Models

**By: Mathieu Blauw**

The goal of this research was to gain insights into the performance of commonly applied parametric car-following models on representing the driving behaviour of ACC systems. Optimal model calibration was obtained by investigating the sensitivity of the model calibration to synthetic data. Investigated were the calibration methodology and the quality and quantity of calibration data. Models are calibrated to real-world driving data from an Audi A4 from 2017. For the ACC system, it was found: 1)the ACC system exhibits non-linear driving behaviour, 2)the acceleration depends on the current velocity and distance to the desired velocity, 3)the system does not consider an intelligent braking strategy and is thus not able of handling safety-critical driving situations and 4)the model includes a sub-controller which ensures comfortabledriving behaviour.

### Modelling traffic in the Randstad using a dynamic zone model based on the Network Fundamental Diagram

**By: Mark Sloot**

This master thesis focusses on the usability of a dynamic zone-based traffic assignment model for large areas,using the concept of the Network Fundamental diagram. These kinds of models differ from the traditional traffic assignment models, because the conventional models are suitable for modelling traffic networks consisting of links, rather than zone. The simulation model that is used in this case study for the Randstad is a newly developed model in-spired by the Network Transmission Model. The main difference lies in the way the demand to the neighboring zone is constructed.

The model then works as follows. For each observed density on the freeway network the corresponding densities on the lower-hierarchy roads are calculated, as well as the weighted average density forthe full network. With these densities and the fundamental diagrams, the weighted average network flow is calculated. The resulting weighted average densities and flows are then plotted in theK,P-diagram and a multi-linear line is fitted through the points. With this input, the simulation can be performed. Generally, the model is able to reproduce the shapes of the outflow patterns, taking into account the de-sired range of 20%, but some zones show large deviations. Especially for the zones Den Haag and Rotterdam the outflow is particularly during morning peak much higher than the observed value.

### New Intersection Control for Conventional and Automated Vehicles without Traffic Lights

**A combination of self-regulation and individual control**

**By: Anna Cristofoli**

Typically, an intersection consists of a number of approaching roads and a crossing area. This thesis aims to develop a novel control strategy that efficiently controls traffic withdifferent penetration rates by relying purely on wireless communication to integrate the control of automated and human drivers. Automated vehicles are considered to be also connected while conventional vehicles are not. The intersection does not need traffic lights sowhile V2I communications can be used to individually control automated vehicle, the motion of conventional vehicles will be indirectly influenced by controlling the speed of automated vehicles. From a design perspective, the solution framework is translated into a bi-level optimization problem with the ultimate goal of computing the acceleration profile (control signal) of controlled “i-platoons” (i.e., platoons lead by a controlled vehicle, possibly followed by non controlled vehicles) that yields the minimal total delay for all i-platoons in the network. The upper level optimization is a combinatorial optimization that aims to find the sequence of i-platoons with the least delay. In the upper level, each sequence is associated with a control signal which generate the least delay for that sequence. This optimization is solved in the lower level, where, given that sequence, the best acceleration profile is computed. he results show that the traffic control strategy is able to improve the efficiency of the intersection under unsaturated conditions . This results is achieved already with low penetration rate of 20%. During saturated conditions,the control strategy shows a performance drop. When uncontrolled i-platoons start queuing on the minor road and there is not enough platoons on the major road to slow down in favor of theminor road, the traffic condition cannot be improved.

### Flow Capacity of Bottlenecks in a Cycle Storage

**By: Dorus Brouwer**

This master thesis intends to give more insight in the flow capacity of the bottlenecks of such facilities for future design of the cycle storages. The research question of this thesis is: what is the capacity and the variation of the flow at capacity of a bicycle check-in, bicycle gate and bicycle stairs, and what is a suitable method to determine these values?The contribution of this thesis is twofold: 1) a method to measure a confidence interval for the capacity and the variation at capacity of a pedestrian bottleneck, and 2) capacity values for three common bottlenecks in cycle storages at station. The main results are shown below:

The method of this thesis can be summarized as follows: for each of the bottlenecks, measurements have been done which produced an event based log of the flow. During the measurements it was registered if there was a queue. Practitioners are recommended to use the method that is described in this research for future capacity measurements. The results of this study allow the designer to test a design with a simple queuing model.

### Using relative flow data to expose the fundamental diagram

**By: Sebastiaan Thoen**

The fundamental diagram (FD) describes the relation between the macroscopic traffic variables speed, flow,and density in stationary and homogenous traffic. The FD is an important input for traffic control measures. Currently, loop detectors installed in the road are most often used to collect data to estimate the FD. Such loop detectors are expensive, can be out of order, and do not provide information when traffic is at standstill. This thesis introduces a methodology to estimate points of flow and density on the FD using only probe vehicles as moving observers. Areas in space-time (xt--areas) are constructed using the trajectories of moving observers as boundaries. The values of flow and density in an xt-area can be calculated with Edie’s definitions. Furthermore, we developed a simpler method, which we call the Corner Points Delta (CDN) method,to estimate flow and density in each xt-area, using the relative flow and average speed of each boundary as input. For (nearly)stationary and homogeneous traffic conditions, which can be identified with our method, the CDN method can estimate points on the FD with high accuracy. It also provides more data about very congested states,improving on traditional methods that use loop detector data. Because it only requires data collected by vehicles with sensing equipment, the FD can be estimated for road sections where loop detectors are out of order or not installed. The paper indicates possible extensions for a wider application of the method.