405 research outputs found
Conclusion : reflections and lessons from the pandemic
No full textThis concluding chapter presents a summary of the research findings in the previous chapters, along with some reflections for each of the five themes of the book and a discussion of necessary future responses (post-pandemic or in the event of a new pandemic) and topics that require further exploration. The pandemic brought into sharp relief pre-existing social disparities and affected vulnerable populations the most. The economic impacts of the pandemic were diverse and varied by geography, but again certain geographies and economic sectors were more buffered from negative outcomes than others. A lesson and a challenge for policymakers is to find ways to understand and reduce these disparities, instead of pushing them under the rug. The impacts on mobility and travel were dramatic as total trips decreased, transit usage fell dramatically, and telecommuting and active modes of transportation increased. Some positive impacts included an improved air quality, a reduced number of traffic crashes, and a proliferation of walking and biking in some neighbourhoods. As cities are slowly recovering from the pandemic, the challenge is to keep the positive impacts but also find ways to help the transit industry rebound from its plunge. Long-term impacts of the pandemic in terms of changing patterns of work and work arrangements, shopping, recreation, and other human activities that will affect travel need additional time and more research to discern
Adaptive averaging in accelerated descent dynamics
No full textFree to read at publisher website\ud
\ud
We study accelerated descent dynamics for constrained convex optimization. This dynamics can be described naturally as a coupling of a dual variable accumulating gradients at a given rate , and a primal variable obtained as the weighted average of the mirrored dual trajectory, with weights . Using a Lyapunov argument, we give sufficient conditions on and to achieve a desired convergence rate. As an example, we show that the replicator dynamics (an example of mirror descent on the simplex) can be accelerated using a simple averaging scheme. We then propose an adaptive averaging heuristic which adaptively computes the weights to speed up the decrease of the Lyapunov function. We provide guarantees on adaptive averaging in continuous-time, prove that it preserves the quadratic convergence rate of accelerated first-order methods in discrete-time, and give numerical experiments to compare it with existing heuristics, such as adaptive restarting. The experiments indicate that adaptive averaging performs at least as well as adaptive restarting, with significant improvements in some cases
Understanding Network Traffic States using Transfer Learning
No full textLarge-scale network traffic analysis is crucial for many transport applications, ranging from estimation and prediction to control and planning. One of the key issues is how to integrate spatial and temporal analyses efficiently. Deep Learning is gaining momentum as a go-to approach for artificial vision, and transfer learning approaches allow to exploit pretrained models and apply them to new domains. In this paper, we encode traffic states as images and use a pretrained deep convolutional neural network as a feature extractor. Experimental results show how the extracted feature vectors cluster naturally into meaningful network traffic states and illustrate how these network states can be used for traffic state prediction.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Transport and Plannin
Design of Haptic Feedback Control for Steer-by-Wire
No full textThis paper illustrates a comparison of different haptic feedback control strategies; primarily focusing on open and closed-loop methods for a Force-Feedback Steer-by-Wire system. Due to shortcomings caused by the feedback motor impedance in the open loop architecture, the tracking performance is deteriorated. Consequently it is shown that the closed-loop solutions provide an improved response within the desired steering excitation range. The closed-loop possibilities, torque and position control, are designed and objectively compared in terms of performance and stability. The controller objectives are inertia compensation and reference tracking. For a given reference, the stability constraint between the controller gains responsible for the two objectives is contrasting in both the methods. Higher bandwidth is achieved for torque controller, whereas the driver arm inertia limits the position control performance. The linear system analysis is supported by the experimental results
Recommended from our members
From Simulation to the Real World: Deep Reinforcement Learning for Training Robust, Wave-Smoothing Policies for Autonomous Vehicles
No full textWith the advent of autonomous vehicles (AVs) comes a broad array of possibilities for control. Looking beyond the immediate wave of research that is focused on training models that can drive safely, this work looks into the future, and aims to develop models for AVs that can achieve more than safe driving. Traffic dynamics are notoriously difficult to model and capture on the micro-level, with behaviors ranging from human-observable to ones we are not aware of, happening every second. Reinforcement learning (RL) is a method which is effective in capturing structure from highly complex, heavy, behavioral data. In this work, we use RL and leverage its ability to understand complex human-driver and traffic dynamics in order to develop policies that are able to not only drive, but drive in a way that can smooth traffic.With the goal of taking these traffic-smoothing algorithms to the real world, the aim of this work takes a path through three parts, from work done purely in simulation to a eventual 100-AV road test. We first explore the concept of using RL as a means of control for wave-smoothing policy control by examining experiments across a variety of traffic scenarios that demonstrate its effectiveness. This portion happens purely in simulation and explores various components of RL design, from environment design to reward shaping. With the goal of deployment always in mind, we also conduct research on how to develop RL policies that are robust enough to survive the transfer from simulation to the real world while sacrificing minimal performance. Lastly, the work comes together to explore the development and deployment of the MegaVanderTest, the deployment of 100 RL-enabled AVs, and to our knowledge, the largest test of AVs designed to smooth traffic
ASPEK HUKUM PELIBATAN MASYARAKAT DALAM PROSES AMDAL JALAN TOL YOGYA-SOLO DI KALURAHAN PURWOMARTANI KAPANEWON KALASAN PADUKUHAN BAYEN
No full textThe Indonesian government has power over the environment and what is in it, so it has the right to carry out development, one of which is the construction of toll roads. The construction of this toll road has an impact on the surrounding environment so an AMDAL or Environmental Impact Statement (EIS) must be made. Article 26 paragraph 2 of the UUPPLH stipulates that in the AMDAL preparation process it is mandatory to involve the community to hear their
opinions. The research method used in this research is an empirical research method. This legal research uses primary data and secondary data. Primary data was obtained from interview with the source person, while the secondary data was obtained from regulations and laws, books, journals, news, the internet and research reports. The author also using qualitative methods by drawing conclusions from the results of interviews with sources and respondents. The results of this study indicate that the Bayen hamlet community affected by the construction of the Yogya-Solo toll road were not involved in the environmental impact analysis (AMDAL) process for the construction of the toll road, there was no information or announcement, and the Sleman Regency Environmental Service did not know the
initiator of the construction of the Yogya-Solo toll road. This is what makes this case contradictory to Article 26 of the UUPPLH, and in the end the Bayen hamlet community was not involved in the environmental impact analysis (AMDAL)
process for the construction of the Yogya-Solo toll road in Purwomartani Village, Kalasan Subdistrict
Recommended from our members
Routing strategies for the reliable and efficient utilization of road networks
No full textThe research presented in this dissertation aims to develop computationally tractable models and algorithms for the reliable and efficient utilization of capacity restricted transportation networks via route selection and demand redistribution, motivated by the fact that traffic congestion in road networks is a major problem in urban communities. Three related topics are considered, 1) route planning with reliability guarantees, 2) system optimal dynamic traffic assignment, and 3) controlling user equilibrium departure times.Route planning can in many practical settings require finding a route that is both fast and reliable. However, in most operational settings, only deterministic shortest paths are considered, and even when the link travel-times are known to be stochastic the common approach is to simply minimize the expected travel-time. This approach does not account for the variance of the travel-time and gives no reliability guarantees. In many cases, travelers have hard deadlines or are willing to sacrifice some extra travel-time for increased travel-time reliability, such as in commercial routing applications where delivery guarantees need to be met and perishables need to be delivered on time. The research presented in this dissertation develops fast computation techniques for the reliable routing problem known as the stochastic on-time arrival (SOTA) problem, which provides a routing strategy that maximizes the probability of arriving at the destination within a fixed time budget. Selfish user optimal routing strategies can, however, lead to very inefficient traffic equilibria in congested traffic networks. This "Price of Anarchy" can be mitigated using system optimal coordinated routing algorithms. The dissertation considers the system optimal dynamic traffic assignment problem when only a subset of the network agents can be centrally coordinated. A road traffic dynamics model is developed based on the Lighthill-Williams-Richards partial differential equation and a corresponding multi-commodity junction solver. Full Lagrangian paths are assumed to be known for the controllable agents, while only the aggregate split ratios are required for the non-controllable (selfish) agents. The resulting non-linear optimal control problem is solved efficiently using the discrete adjoint method. Spill-back from under-capacitated off-ramps is one of the major causes of congestion during the morning commute. This spill-back induces a capacity drop on the freeway, which then creates a bottleneck for the mainline traffic that is passing by the off-ramp. Therefore, influencing the flow distribution of the vehicles that exit the freeway at the off-ramp can improve the throughput of freeway vehicles that pass this junction. The dissertation studies the generalized morning commute problem where vehicles exiting the freeway at the under-capacitated off-ramp have a fixed desired arrival time and a corresponding equilibrium departure time schedule, and presents strategies to manipulated this equilibrium to maximize throughput on the freeway via incentives or tolls
Recommended from our members
From sim-to-real: learning and deploying autonomous vehicle controllers that improve transportation metrics
No full textThe recent wide availability of semi-autonomous vehicles with distance and lane keep capabilities have created an exciting opportunity to improve the throughput and energy efficiency of the highway by deploying modified control strategies. However, even at current penetration rates, the optimal mechanism for the design of these decentralized, cooperative strategies is an open problem. In this work, we use Multi-Agent Reinforcement Learning (MARL) to investigate, design, and deploy cooperative autonomous vehicles (CAVs) to achieve these goals and demonstrate a field deployment of an RL-based traffic smoothing controller.
We focus on multi-agent reinforcement learning as a mechanism for handling the complexity and non-linearity of large-scale traffic. We start by constructing a standardized suite of benchmark tasks for evaluating the efficacy of learning algorithms in designing controllers for CAVs; we evaluate these algorithms in the centralized setting where all CAVs are actuated by a single controller. We then extend one of these benchmarks, regulation of the inflow to a bottleneck via decentralized CAVs, to the multi-agent setting. We demonstrate that from both low to high penetration rates, CAVs are capable of improving the throughput of a scaled model of the San Francisco-Oakland Bay Bridge and investigate challenges in scaling our methods in open-network settings where vehicles can enter and exit the system.
In preparation for a road test intended to demonstrate stop-and-go wave smoothing on large scale networks, we next study energy optimization of a full-scale model of a section of the I-210 in Los Angeles. Using Proximal Policy Optimization with an augmented value function we demonstrate that we are able to sharply improve the miles-per-gallon of the system and that the resultant controller is robust to likely variations of the system such as system speed and CAV penetration rate. However, we observe that the resultant waves are very unrealistic and additional calibration using higher resolution data is needed.
With the goal of designing a more calibrated simulator, we pursue two approaches: one approach focuses on designing new driver models using available data-sets from Waymo and another approach focused on the use of collected data from the field deployment site. In the first approach, we design a new simulator that 1) efficiently represents the partially observable view-cone of human drivers and investigate whether learning safe driving policies in the simulator yields human-like behavior 2) serves as a challenging MARL benchmark. We observe promising signs of human-similarity from agents trained in the simulator. In the more direct approach, we collect data from the deployment site and use it to design a new, simplified simulator capable of using the collected data while maintaining a high simulation speed. We design energy-improving CAVs in this simulator and demonstrate that these CAVs can be successfully and safely used in a field deployment test
Recommended from our members
On learning Game-Theoretical models with Application to Urban Mobility
No full textModeling real-world processes as convex optimization or variational inequality problems is a common practice as it enables to leverage powerful mathematical tools for the study of such processes. For example, in economics, knowing the consumer utility function enables to adjust prices to achieve some demand level. In control, a low complexity controller requires less computation for little performance loss. In transportation science, the selfish behavior of agents (from shorted path routing) leads to an aggregate cost in the network worse than the system’s optimum, and which can be analytically quantified. Taxation schemes can be designed to incentivize system optimal drivers’ decisions.In the first part of our work, we briefly review fundamental results in convex optimization, variational inequality theory, and game theory. We also focus on the selfish routing game, which is a popular game-theoretical framework to model the urban transportation network. In particular, we study the impact of the increasing penetration of routing apps on road usage. Its conclusions apply both to manned vehicles in which human drivers follow app directions, and unmanned vehicles following shortest path algorithms. To address the problem caused by the increased usage of routing apps, we model two distinct classes of users, one having limited knowledge of low-capacity road links. This approach is in sharp contrast with some previous studies assuming that each user has full knowledge of the network and optimizes his/her own travel time. We show that the increased usage of GPS routing provides a lot of benefits on the road network of Los Angeles, such as decrease in average travel times and total vehicle miles traveled. However, this global increased efficiency in urban mobility has negative impacts as well, which are not addressed by the scientific community: increase in traffic in cities bordering highway from users taking local routes to avoid congestion.In the second part, we explore the ability of low complexity game-theoretical models to accurately approximate real transportation systems. For example, system mischaracterizations in selfish routing can cause taxes designed for one problem instance to incentivize inefficient behavior on different, yet closely-related instances. Hence, we want to be able to measure the quality of the learned model. In the present work, we present a statistical framework for the fitting of equilibrium models based on measurements of edge flows using the (standard) empirical risk minimization principle, by choosing the fit giving the lowest expected loss (the distance between the observed and predicted outputs) under the empirical measure. Hence, for the class of models of interest, it is critical to be able to have theoretical guarantees on the quality of the fit. We then present a computational methodology for imputing the map of an equilibrium model, and propose a statistical hypothesis test for validating the trained model against the true one.In the third part, we explore existing work for estimating link and route flows, and we propose two novel frameworks for traffic estimation. In the first framework, we focus on estimating the highway traffic, which is modeled as a discretized hyperbolic scalar partial differential equations. The system is written as a switching dynamical system, with a state space partitioned into an exponential number of polyhedra in which one mode is active. We propose a feasible approach based on the interactive multiple model (IMM), and apply the k-means algorithm on historical data to partition modes into clusters, thus reducing the number of modes. In the second framework, we develop a convex optimization methodology for the route flow estimation problem from the fusion of vehicle count and cellular network data. The proposed approach is versatile: it is compatible with other data sources, and it is model agnostic and thus compatible with user equilibrium, system-optimum, Stackelberg concepts, and other models. The framework is validated on the I-210 corridor near Los Angeles, where we achieve 90% route flow accuracy with 1033 traffic sensors and 1000 cellular towers covering a large network of highways and arterials with more than 20,000 links
- …
