Your Selections

Carnegie Mellon University
Show Only

Collections

File Formats

Content Types

Dates

Sectors

Topics

Authors

Publishers

Affiliations

Events

   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Machine Learning with Decision Trees and Multi-Armed Bandits: An Interactive Vehicle Recommender System

Carnegie Mellon University-Tong Yu, Ole Mengshoel
Ford Motor Co., Ltd.-Dominique Meroux, Zhen Jiang
Published 2019-04-02 by SAE International in United States
Recommender systems guide a user to useful objects in a large space of possible options in a personalized way. In this paper, we study recommender systems for vehicles. Compared to previous research on recommender systems in other domains (e.g., movies or music), there are two major challenges associated with recommending vehicles. First, typical customers purchase fewer cars than movies or pieces of music. Thus, it is difficult to obtain rich information about a customer’s vehicle purchase history. Second, content information obtained about a customer (e.g., demographics, vehicle preferences, etc.) is also difficult to acquire during a relatively short stay in a dealership. To address these two challenges, we propose an interactive vehicle recommender system based a novel machine learning method that integrates decision trees and multi-armed bandits. Decision tree learning effectively selects important questions to ask the customer and encodes the customer's key preferences. With these preferences as prior information, the multi-armed bandit algorithm, using Thompson sampling, efficiently leverages the user’s feedback to improve the recommendations in an online fashion. The empirical results show that…
This content contains downloadable datasets
Annotation ability available
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Safety Argument Considerations for Public Road Testing of Autonomous Vehicles

SAE International Journal of Advances and Current Practices in Mobility

Carnegie Mellon University-Philip Koopman
Edge Case Research-Beth Osyk
  • Journal Article
  • 2019-01-0123
Published 2019-04-02 by SAE International in United States
Autonomous vehicle (AV) developers test extensively on public roads, potentially putting other road users at risk. A safety case for human supervision of road testing could improve safety transparency. A credible safety case should include: (1) the supervisor must be alert and able to respond to an autonomy failure in a timely manner, (2) the supervisor must adequately manage autonomy failures, and (3) the autonomy failure profile must be compatible with effective human supervision.Human supervisors and autonomous test vehicles form a combined human-autonomy system, with the total rate of observed failures including the product of the autonomy failure rate and the rate of unsuccessful failure mitigation by the supervisor. A difficulty is that human ability varies in a nonlinear way with autonomy failure rates, counter-intuitively making it more difficult for a supervisor to assure safety as autonomy maturity improves. Thus, road testing safety cases must account for both the expected failures during testing and the practical effectiveness of human supervisors given that failure profile. This paper outlines a high level safety case that identifies key…
Annotation ability available
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Automatic Hex-Dominant Mesh Generation for Complex Flow Configurations

SAE International Journal of Engines

Carnegie Mellon University-Soji Yamakawa, Satbir Singh, Kenji Shimada
University of Wisconsin-Nihar Sawant
  • Journal Article
  • 2018-01-0477
Published 2018-04-03 by SAE International in United States
A method for automatically generating hex-dominant meshes for Computational Fluid Dynamics (CFD) applications is presented in this article. Two important regions of the mesh for any CFD simulation are the interior mesh and the boundary layer mesh. The interior mesh needs to be fine in the critical flow regions to ensure accurate solutions. The proposed method uses Bubble Mesh algorithm which packs bubbles inside the geometry to generate the mesh nodes. Algorithm was tested for sample flow problems and improvements were made to interior and boundary layer mesh generation methods. The interior mesh is generated using directionality and sizing control functions specified on the points of a 3D grid generated over the entire geometry. This offers a flexible control over mesh sizing and local mesh refinement. Boundary layer mesh is important to accurately model the physics of boundary layer near the geometry walls. In the proposed method, boundary elements in the mesh are split into multiple divisions with the first division having the smallest thickness to ensure it lies inside the physical boundary layer. The…
This content contains downloadable datasets
Annotation ability available
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Influence of Discretization Schemes and LES Subgrid Models on Flow Field Predictions for a Motored Optical Engine

SAE International Journal of Engines

Carnegie Mellon University-Varun Haresh Nichani, Roberto Jaime, Satbir Singh
General Motors LLC-Xiaofeng Yang
  • Journal Article
  • 2018-01-0185
Published 2018-04-03 by SAE International in United States
Large-eddy simulations (LES) of a motoring single-cylinder engine with transparent combustion chamber (TCC-II) are carried out using a commercially available computer code, CONVERGE. Numerical predictions are compared with high-speed particle image velocimetry (PIV) measurements. Predictions of two spatial discretization schemes, namely, numerically stabilized central difference scheme (CDS) and fully upwind scheme are compared. Four different subgrid scale (SGS) models; a non-eddy viscosity dynamic structure turbulence (DST) model of Pomraning and Rutland, one-equation eddy-viscosity (1-Eqn) model of Menon et al., a zeroequation eddy-viscosity model of Vreman, and the zeroequation standard Smagorinsky model are employed on two different grid configurations. Additionally, simulations are also performed by deactivating the LES SGS models. It is found that the predictions when using the numerically stabilized CDS are significantly better than using the fully upwind scheme. The LES SGS models clearly make an impact on predicted flow field although the impact is not always positive. Overall, the eddy-viscosity model by Vreman provided the best predictions of flow statistics compared to the other LES models used in this work. A parameter called…
This content contains downloadable datasets
Annotation ability available
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Challenges in Autonomous Vehicle Testing and Validation

SAE International Journal of Transportation Safety

Carnegie Mellon University-Philip Koopman
Edge Case Research LLC-Michael Wagner
  • Journal Article
  • 2016-01-0128
Published 2016-04-05 by SAE International in United States
Software testing is all too often simply a bug hunt rather than a well-considered exercise in ensuring quality. A more methodical approach than a simple cycle of system-level test-fail-patch-test will be required to deploy safe autonomous vehicles at scale. The ISO 26262 development V process sets up a framework that ties each type of testing to a corresponding design or requirement document, but presents challenges when adapted to deal with the sorts of novel testing problems that face autonomous vehicles. This paper identifies five major challenge areas in testing according to the V model for autonomous vehicles: driver out of the loop, complex requirements, non-deterministic algorithms, inductive learning algorithms, and fail-operational systems. General solution approaches that seem promising across these different challenge areas include: phased deployment using successively relaxed operational scenarios, use of a monitor/actuator pair architecture to separate the most complex autonomy functions from simpler safety functions, and fault injection as a way to perform more efficient edge case testing. While significant challenges remain in safety-certifying the type of algorithms that provide high-level autonomy…
Annotation ability available
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Intersection Management using Vehicular Networks

Carnegie Mellon University-Reza Azimi, Gaurav Bhatia, Raj Rajkumar
General Motors Company-Priyantha Mudalige
Published 2012-04-16 by SAE International in United States
Driving through intersections can be potentially dangerous because nearly 23 percent of the total automotive related fatalities and almost 1 million injury-causing crashes occur at or within intersections every year [1]. The impact of traffic intersections on trip delays also leads to waste of human and natural resources. Our goal is to increase the safety and throughput of traffic intersections using co-operative driving.In earlier work [2], we have proposed a family of vehicular network protocols, which use Dedicated Short Range Communications (DSRC) and Wireless Access in Vehicular Environment (WAVE) technologies to manage a vehicle's movement at intersections Specifically, we have provided a collision detection algorithm at intersections (CDAI) to avoid potential crashes at or near intersections and improve safety. We have shown that vehicle-to-vehicle (V2V) communications can be used to significantly decrease the trip delays introduced by traffic lights and stop signs. In this paper, we investigate the use of more realistic controller models and higher concurrency to improve V2V intersection protocols for autonomous driving at intersections. We quantify the throughput enhancements due to the…
Annotation ability available
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

AUTOSAR Extensions for Predictable Task Synchronization in Multi-Core ECUs

Carnegie Mellon University-Karthik Singaram Lakshmanan, Gaurav Bhatia, Ragunathan Rajkumar
Published 2011-04-12 by SAE International in United States
Multi-core processors are becoming increasingly prevalent, with several multi-core solutions being offered for the automotive sector. Recognizing this trend, the AUTomotive Open System ARchitecture (AUTOSAR) standard Version 4.0 has introduced support for multi-core embedded real-time operating systems. A key element of the AUTOSAR multi-core specification is the spinlock mechanism for inter-core task synchronization. In this paper, we study this spinlock mechanism from the standpoint of timing predictability. We describe the timing uncertainties introduced by standard test-and-set spinlock mechanisms, and provide a predictable priority-driven solution for inter-core task synchronization.The proposed solution is to arbitrate critical sections using the well-established Multi-processor Priority Ceiling Protocol [3], which is the multiprocessor version of the ceiling protocol for uniprocessors [1, 2] used by AUTOSAR. We also present the associated analysis that can be used in conjunction with the AUTOSAR task model to bound the worst-case waiting times for accessing shared resources. The timing predictability provided by our protocol is an important requirement for automotive applications from both certification and validation standpoints.
Annotation ability available
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Vehicular Networks for Collision Avoidance at Intersections

SAE International Journal of Passenger Cars - Mechanical Systems

Carnegie Mellon University-Seyed Reza Azimi, Gaurav Bhatia, Ragunathan (Raj) Rajkumar
GM Technical Center-Priyantha Mudalige
  • Journal Article
  • 2011-01-0573
Published 2011-04-12 by SAE International in United States
A substantial fraction of automotive collisions occur at intersections. Statistics collected by the Federal Highway Administration (FHWA) show that more than 2.8 million intersection-related crashes occur in the United States each year, with such crashes constituting more than 44 percent of all reported crashes [12]. In addition, there is a desire to increase throughput at intersections by reducing the delay introduced by stop signs and traffic signals. In the future, when dealing with autonomous vehicles, some form of co-operative driving is also necessary at intersections to address safety and throughput concerns.In this paper, we investigate the use of vehicle-to-vehicle (V2V) communications to enable the navigation of traffic intersections, to mitigate collision risks, and to increase intersection throughput significantly. Specifically, we design a vehicular network protocol that integrates with mobile wireless radio communication standards such as Dedicated Short Range Communications (DSRC) and Wireless Access in a Vehicular Environment (WAVE). This protocol relies primarily on using V2V communications, GPS and other automotive sensors to safely navigate intersections and also to enable autonomous vehicle control. Vehicles use DSRC/WAVE…
Annotation ability available
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

New Model for University Research and Technology Transfer

SAE International Journal of Passenger Cars - Electronic and Electrical Systems

Carnegie Mellon University-Stephen A. DiAntonio
  • Journal Article
  • 2010-01-2301
Published 2010-10-19 by SAE International in United States
As non-profit organizations, universities serve the public good through their primary missions of teaching and conducting fundamental research. In the past, American universities obtained the bulk of their funding in the form of basic research grants from U.S. government agencies such as NSF, NIH, USDA, DOE and the DoD. These grants funded graduate students and their faculty advisors who published their research results to benefit society through the free dissemination of knowledge.Though the academic research model still adequately describes most university research programs, a new model has emerged that can accommodate new requirements imposed by government agencies and can provide R&D services for commercial companies. The model serves sponsors who want to see university technology applied in settings directly relevant to their mission or business operations. This allows the sponsors to see “what's possible” with the technology without making a large investment to build or acquire an internal development capability. Those applications that show operational viability get transferred to the sponsor by imparting know how to sponsor employees, licensing, and providing well-documented technical data packages.One…
Annotation ability available
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Agent-Based Modeling and Simulation of Collaborative Air Traffic Flow Management using Brahms

SAE International Journal of Aerospace

Carnegie Mellon University-Maarten Sierhuis
Craig Technologies-Peter A. Jarvis
  • Journal Article
  • 2009-01-3202
Published 2009-11-10 by SAE International in United States
The air traffic demand on the US national airspace frequently exceeds its available capacity. In current operations, the Air Traffic Service Provider designs and implements air traffic management initiatives with minimal interaction with the airlines. NASA and its partners have developed a new collaborative air traffic flow management concept of operations that involves the users of the airspace to a greater degree. In this paper, we describe an agent-based simulation of the new concept of operations and our planned experimentation to determine if the new concept of operations will lead to better utilization of the national airspace.
Annotation ability available