Your Selections

Cartography
Show Only

Collections

File Formats

Content Types

Dates

Sectors

Topics

Authors

Publishers

Affiliations

Committees

Events

Magazine

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

Autonomous Vehicle Multi-Sensors Localization in Unstructured Environment

FEV North America Inc.-Qusay Alrousan, Hamzeh Alzu'bi, Andrew Pfeil, Tom Tasky
  • Technical Paper
  • 2020-01-1029
To be published on 2020-04-14 by SAE International in United States
Autonomous driving in unstructured environments is a significant challenge due to the inconsistency of important information for localization such as lane markings. To reduce the uncertainty of vehicle localization in such environments, sensor fusion of LiDAR, Radar, Camera, GPS/IMU, and Odometry sensors is utilized. This paper discusses a hybrid localization technique developed using: LiDAR based Simultaneous Localization and Mapping (SLAM), GPS/IMU and Odometry data, and object lists from Radar and Camera sensors. An Extended Kalman Filter (EKF) is utilized to fuse data from all sensors in two phases. In the preliminary stage, the SLAM-based vehicle coordinates are fused with the GPS-based positioning. The output of this stage is then fused with the objects-based localization. This approach was successfully tested on FEV’s Smart Vehicle Demonstrator at FEV’s HQ representing a complicated test environment with dynamic and static objects. The test results show that multi-sensor fusion improves the vehicle’s localization compared to GPS or LiDAR alone.
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

A Stability-Guaranteed Time-Delay Range for Feedback Control of Autonomous Vehicles

Yildiz Technical University-Ahmet Kirli, Mehmet Selçuk Arslan
  • Technical Paper
  • 2020-01-0090
To be published on 2020-04-14 by SAE International in United States
The vehicles with level-5 autonomy (L5AVs) have no human driver in the loop are also known as self-driving cars. L5AVs are assumed the next generation of ground transportation, which have growing attention from both industry and academia in most recent years. Most of the work related to feedback strategies of L5AVs are on developing mapping systems through a variety of sensors. These systems can be considered as an analogue to the perception and central nervous system of human drivers. For instance, innovative visualization systems are more powerful when compared to the visual perception system of a person, yet, mapping demands high computation loads. This burden causes delay in the feedback loop and thus, it might have an unfavorable influence on proper and safe control action. This study investigates the effect of time delay occurring in mapping systems on the stability of the controlled vehicle. An algorithm entitled as “Cluster Treatment of Characteristic Roots - CTCR” is used to calculate a safe delay range as a remedy for the time delay caused by mapping systems. The…
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Advancements of MEMristor Materials in Neuromorphic Computing for Autonomous Systems.

Wayne State University-Kyle W. Brown
  • Technical Paper
  • 2020-01-0088
To be published on 2020-04-14 by SAE International in United States
The advancements in analog electronics has spurred the development of neuromorphic computing which can replicate bio-neurological processes using artificial synapses. Artificial synapses can process information faster and more efficiently than CPUs for specialized applications like sparse coding, graph searches, and constraint-satisfaction problems. Neuromorphic systems offset CPU’s lack of processing power to solve complex tasks and computations, higher parallelism, novel neural-inspired algorithms, and optimizations. Neural-inspired algorithms such as sparse coding, simultaneous localization and mapping (SLAM), path planning, and object tracking event-based cameras are necessary in development of autonomous systems. As the industry and academia realizes the limitations posed Moore’s Law, new computing and performance by MEMristors has enabled continued process-node scaling. New technology like Intel’s inspired neuromorphic microchip demonstrates the benefits of a specialized architecture for emerging applications, including some of the computational problems hardest for the internet of things (IoT) and autonomous devices to support. As new complex computing workloads grow the need for specialized architectures designed for specific applications will be in demand. Specialized architectures using specific applications are ideal for real-world applications, from…
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

N&V Component Structural Integration and Mounted Component Durability Implications

General Motors LLC-Mark Stebbins, Joseph Schudt
  • Technical Paper
  • 2020-01-1396
To be published on 2020-04-14 by SAE International in United States
Exterior component integration has presented competing integration challenges for suitable exterior styling, safety, N&V structural feel and component durability balanced performances. Industry standard practice of N&V vehicle mode mapping uses vehicle source, path and receiver considerations for component mode frequency placement. Mounted component mode frequency performance can have an influence on structural feel and durability performances. Component integration headwinds have increased with additional nonstructural component styling content, component size, component mass and added sensor modules. Based on first principles, the effective mass of exterior mounted components are increased due to the geometric overhang from structure. Component input vibration levels are compounded due to the cantilevered nature as well as relative positioning to the suspension and propulsion vehicle source inputs locations. Examples of vehicle end mounted components include but are not limited to headlamps, side mirrors, end gates, bumpers and fascia. Our goal is to establish basic expectations for the behavior of these systems, and ultimately to consolidate our existing rational and approaches that can be applied to such integrated systems. The focus of this paper…
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Real-time Motion Classification of LiDAR Point Detection for Automated Vehicles

Hanyang University-Chansoo Kim, Sungjin Cho, Myoungho Sunwoo
Konkuk University-Kichun Jo
  • Technical Paper
  • 2020-01-0703
To be published on 2020-04-14 by SAE International in United States
A Light Detection And Ranging (LiDAR) is now becoming an essential sensor for an autonomous vehicle. The LiDAR provides the surrounding environment information of the vehicle in the form of a point cloud. A decision-making system of the autonomous car is able to determine a safe and comfort maneuver by utilizing the detected LiDAR point cloud. If the movement class (dynamic or static) of detected points can be provided by LiDAR, the decision-making system is able to plan the appropriate motion of the autonomous vehicle according to the movement of the object. This paper proposes a real-time process to segment the motion states of LiDAR points. The basic principle of the classification algorithm is to classify the point-wise movement of a target point cloud through the other point clouds and sensor poses. First, a fixed-size buffer store the LiDAR point clouds and sensor poses for a constant time window. Second, motion beliefs of the target point cloud against other point clouds and sensor pose in the buffer are estimated, respectively. Each motion belief of the…
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Using Reinforcement Learning and Simulation to Develop Autonomous Vehicle Control Strategies

Amazon Web Services-Sahika Genc, Premkumar Rangarajan
PolySync Technologies-Anthony Navarro
  • Technical Paper
  • 2020-01-0737
To be published on 2020-04-14 by SAE International in United States
While machine learning in autonomous vehicles development has increased significantly in the past few years, the use of reinforcement learning (RL) methods has only recently been applied. Convolutional neural networks (CNNs) became common for their powerful object detection and identification and even provided end-to-end control of an autonomous vehicle. However, one of the requirements of a CNN is a large amount of labeled data to inform the neural network. While data is becoming more accessible, these networks are still sensitive to the format and collection environment which makes the use of others’ data more difficult. In contrast, RL develops solutions in a simulation environment by trial and error without labeled data. Our research expands upon previous research in RL and proximal policy optimization (PPO) and the application of these algorithms to 1/18th scale cars by expanding the application of this control strategy to a full-sized passenger vehicle. By using this method of unsupervised learning, our research demonstrates the ability to learn new control strategies while in a simulation environment without the need for large amounts…
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

A Method for Mapping a Light Source Utilizing HDR Imagery

JS Forensic Consulting, LLC-Jeffrey Suway
Momenta, LLC-Anthony Cornetto
  • Technical Paper
  • 2020-01-0566
To be published on 2020-04-14 by SAE International in United States
Mapping a light source, any light source, is of broad interest to accident reconstructionists, human factors professionals and lighting experts. Such mappings are useful for a variety of purposes, including determining the effectiveness and appropriateness of lighting installations, and performing visibility analyses for accident case studies. Currently, mapping a light source can be achieved with several different methods. One such method is to use an illuminance meter and physically measure each point of interest on the roadway. Another method utilizes a goniometer to measure the luminous intensity distribution, this is a near-field measurement. Both methods require significant time and the goniometric method requires extensive equipment in a lab. A third method measures illumination distribution in the far-field using a colorimeter or photometer. These systems utilize a CCD sensor to measure the illuminance distribution and then software can convert that illuminance distribution to an IES file for use in a Physically-Based Rendering (PBR) engine. Again, this photometer method requires extensive equipment and the measurements must be taken in a laboratory setting. The method presented in this…
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Optical Characterization of the Combustion Process Inside a Large-Bore Fual-Fuel Two-Stroke Marine Engine by Using Multiple High-Speed Cameras

Lund University-Alexios Matamis, Mattias Richter
MAN Energy Solutions-Johan Hult, Eric Baudoin, Stefan Mayer
  • Technical Paper
  • 2020-01-0788
To be published on 2020-04-14 by SAE International in United States
Dual-fuel engines for marine propulsion are gaining in importance due to operational and environmental benefits. Here the combustion in a dual-fuel marine engine operating on diesel and natural gas is studied using a multiple high-speed camera arrangement. By recording the natural flame emission from three different directions the flame position inside the engine cylinder can be spatially mapped and tracked in time. Through space carving a rough estimate of the three-dimensional (3D) flame contour can be obtained. From this contour properties like flame length and height, as well as ignition locations can be extracted. The multi-camera imaging is applied to a dual-fuel marine two-stroke engine with a bore diameter of 0.5 m and a stroke of 2.2 m. Both liquid and gaseous fuels are directly injected at high pressure, using separate injection systems. Optical access is obtained using borescope inserts, resulting in a minimum disturbance to the cylinder geometry. In this type of engine, with fuel injection from positions at the rim of the cylinder, the flame morphology becomes asymmetric. The optical spatial mapping and…
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Road Curvature Decomposition for Autonomous Guidance

University of Nebraska-Lincoln-Ricardo Jacome, Cody Stolle, Michael Sweigard
  • Technical Paper
  • 2020-01-1024
To be published on 2020-04-14 by SAE International in United States
Vehicle autonomy is critically dependent on an accurate identification and mathematical representation of road and lane geometries. Many road lane identification systems are ad hoc (e.g., machine vision and lane keeping systems) or rely on polynomial approximations of road data and GPS positioning. A novel system is proposed in which geodetic road data is parsed along road directions and digitally stored in a road data matrix. Using mapping algorithms, the road data is converted to a smooth, differentiable path which connects critical road coordinates with curvature vectors and changes to road tangent angles. Different road data sources such as GPS or geographical scans were evaluated with this method and compared to current road design standards as per the American Association of State Highway and Transportation Officials. This approach takes advantage of standard roadway design practices, which rely on speed limit, superelevation, and empirical data for maximum lateral acceleration tolerance to determine acceptable radii of curvature for different classes of roadways. Successful implementation of this technology could accelerate autonomous vehicle and navigation research and development for…
   This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Automotive Dimensional Quality Control with Geometry Tree Process

FCA US LLC-James Cole, Yuqin Wang, Robert Bertucci
  • Technical Paper
  • 2020-01-0480
To be published on 2020-04-14 by SAE International in United States
Geometry Tree is a term describing the product assembly structure and the manufacturing process for the product. The concept refers to the assembly structure of the final vehicle (the Part Tree) and the assembly process and tools for the final product (the Process Tree). In the past few years, the Geometry Tree-based quality process was piloted in the FCA NAFTA region and has since evolved into a standardized quality control process. In the Part Tree process, the coordinated measurements and naming convention are enforced throughout the different levels of product sub-assemblies and measurement processes. The Process Tree, on the other hand, includes both prominently identified assembly tools and the mapping of key product characteristics to key assembly tools. The benefits of directly tying critical customer characteristics to actual machine components that have a high propensity to influence them is both preventive and reactive. This article describes the integrated Geometry Tree quality process and how it has been implemented at the FCA vehicle assembly plants and in the dimensional data management system.