Matching Items (3)
Filtering by
- All Subjects: artificial intelligence
Description
In this dissertation, a cyber-physical system called MIDAS (Managing Interacting Demand And Supply) has been developed, where the “supply” refers to the transportation infrastructure including traffic controls while the “demand” refers to its dynamic traffic loads. The strength of MIDAS lies in its ability to proactively control and manage mixed vehicular traffic, having various levels of autonomy, through traffic intersections. Using real-time traffic control algorithms MIDAS minimizes wait times, congestion, and travel times on existing roadways. For traffic engineers, efficient control of complicated traffic movements used at diamond interchanges (DI), which interface streets with freeways, is challenging for normal human driven vehicular traffic, let alone for communicationally-connected vehicles (CVs) due to stochastic demand and uncertainties. This dissertation first develops a proactive traffic control algorithm, MIDAS, using forward-recursion dynamic programming (DP), for scheduling large set of traffic movements of non-connected vehicles and CVs at the DIs, over a finite-time horizon. MIDAS captures measurements from fixed detectors and captures Lagrangian measurements from CVs, to estimate link travel times, arrival times and turning movements. Simulation study shows MIDAS’ outperforms (a) a current optimal state-of-art optimal fixed-cycle time control scheme, and (b) a state-of-art traffic adaptive cycle-free scheme.
Subsequently, this dissertation addresses the challenges of improving the road capacity by platooning fully autonomous vehicles (AVs), resulting in smaller headways and greater road utilization. With the MIDAS AI (Autonomous Intersection) control, an effective platooning strategy is developed, and optimal release sequence of AVs is determined using a new forward-recursive DP that minimizes the time-loss delays of AVs. MIDAS AI evaluates the DP decisions every second and communicates optimal actions to the AVs.
Although MIDAS AI’s exact DP achieves optimal solution in almost real-time compared to other exact algorithms, it suffers from scalability. To address this challenge, the dissertation then develops MIDAS RAIC (Reinforced Autonomous Intersection Control), a deep reinforcement learning based real-time dynamic traffic control system for AVs at an intersection. Simulation results show the proposed deep Q-learning architecture trains MIDAS RAIC to learn a near-optimal policy that minimizes the total cumulative time loss delay and performs nearly as well as the MIDAS AI.
ContributorsPotluri, Viswanath (Author) / Mirchandani, Pitu (Thesis advisor) / Ju, Feng (Committee member) / Zhou, Xuesong (Committee member) / Sefair, Jorge (Committee member) / Arizona State University (Publisher)
Created2023
Description
The Cognitive Decision Support (CDS) model is proposed. The model is widely applicable and scales to realistic, complex decision problems based on adaptive learning. The utility of a decision is discussed and four types of decisions associated with CDS model are identified. The CDS model is designed to learn decision utilities. Data enrichment is introduced to promote the effectiveness of learning. Grouping is introduced for large-scale decision learning. Introspection and adjustment are presented for adaptive learning. Triage recommendation is incorporated to indicate the trustworthiness of suggested decisions.
The CDS model and methodologies are integrated into an architecture using concepts from cognitive computing. The proposed architecture is implemented with an example use case to inventory management.
Reinforcement learning (RL) is discussed as an alternative, generalized adaptive learning engine for the CDS system to handle the complexity of many problems with unknown environments. An adaptive state dimension with context that can increase with newly available information is discussed. Several enhanced components for RL which are critical for complex use cases are integrated. Deep Q networks are embedded with the adaptive learning methodologies and applied to an example supply chain management problem on capacity planning.
A new approach using Ito stochastic processes is proposed as a more generalized method to generate non-stationary demands in various patterns that can be used in decision problems. The proposed method generates demands with varying non-stationary patterns, including trend, cyclical, seasonal, and irregular patterns. Conventional approaches are identified as special cases of the proposed method. Demands are illustrated in realistic settings for various decision models. Various statistical criteria are applied to filter the generated demands. The method is applied to a real-world example.
The CDS model and methodologies are integrated into an architecture using concepts from cognitive computing. The proposed architecture is implemented with an example use case to inventory management.
Reinforcement learning (RL) is discussed as an alternative, generalized adaptive learning engine for the CDS system to handle the complexity of many problems with unknown environments. An adaptive state dimension with context that can increase with newly available information is discussed. Several enhanced components for RL which are critical for complex use cases are integrated. Deep Q networks are embedded with the adaptive learning methodologies and applied to an example supply chain management problem on capacity planning.
A new approach using Ito stochastic processes is proposed as a more generalized method to generate non-stationary demands in various patterns that can be used in decision problems. The proposed method generates demands with varying non-stationary patterns, including trend, cyclical, seasonal, and irregular patterns. Conventional approaches are identified as special cases of the proposed method. Demands are illustrated in realistic settings for various decision models. Various statistical criteria are applied to filter the generated demands. The method is applied to a real-world example.
ContributorsKee, Seho (Author) / Runger, George C. (Thesis advisor) / Escobedo, Adolfo (Committee member) / Gel, Esma (Committee member) / Janakiram, Mani (Committee member) / Rogers, Dale (Committee member) / Arizona State University (Publisher)
Created2020
Description
Collecting accurate collective decisions via crowdsourcing
is challenging due to cognitive biases, varying
worker expertise, and varying subjective scales. This
work investigates new ways to determine collective decisions
by prompting users to provide input in multiple
formats. A crowdsourced task is created that aims
to determine ground-truth by collecting information in
two different ways: rankings and numerical estimates.
Results indicate that accurate collective decisions can
be achieved with less people when ordinal and cardinal
information is collected and aggregated together
using consensus-based, multimodal models. We also
show that presenting users with larger problems produces
more valuable ordinal information, and is a more
efficient way to collect an aggregate ranking. As a result,
we suggest input-elicitation to be more widely considered
for future work in crowdsourcing and incorporated
into future platforms to improve accuracy and efficiency.
is challenging due to cognitive biases, varying
worker expertise, and varying subjective scales. This
work investigates new ways to determine collective decisions
by prompting users to provide input in multiple
formats. A crowdsourced task is created that aims
to determine ground-truth by collecting information in
two different ways: rankings and numerical estimates.
Results indicate that accurate collective decisions can
be achieved with less people when ordinal and cardinal
information is collected and aggregated together
using consensus-based, multimodal models. We also
show that presenting users with larger problems produces
more valuable ordinal information, and is a more
efficient way to collect an aggregate ranking. As a result,
we suggest input-elicitation to be more widely considered
for future work in crowdsourcing and incorporated
into future platforms to improve accuracy and efficiency.
ContributorsKemmer, Ryan Wyeth (Author) / Escobedo, Adolfo (Thesis director) / Maciejewski, Ross (Committee member) / Computing and Informatics Program (Contributor) / Computer Science and Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2020-05