Search Results

The Ohio State University Center for Intelligent Transportation Research (CITR) has developed three automated vehicles demonstrating advanced cruise control, automated steering control for lane keeping, and autonomous behavior including automated stopping and lane changes in reaction to other vehicles. Various sensors were used, including a radar reflective stripe system and a vision based system for lane position sensing, a radar system and a scanning laser rangefinding system for the detection of objects ahead of the vehicle, and various supporting sensors including side looking radars and an angular rate gyroscope. These vehicles were demonstrated at the National Automated Highway System Consortium (NAHSC) 1997 Technical Feasibility Demonstration in a scenario involving mixed autonomous and manually driven vehicles. This paper describes the demonstration, the vehicle sensing, control, and computational hardware, and the vehicle control software.

With increasing demands to reduce emissions from internal combustion engines, engine manufacturers are forced to seek out new technology. One such technology employed primarily in the diesel and two-stroke engine community is direct-injection (DI). Direct injection has shown promising results in reduction of CO and NOx for both two- and four-stroke engines. While having been used for several years in the diesel industry, direct injection has been scrutinized for an inability to meet future requirements to reduce particulate matter emissions. Direct injection has also came under fire for complicating fuel delivery systems, thus making it cost prohibitive for small utility engine manufacturers. Recent research shows that the application of piezo-driven actuators has a positive effect on soot formation reduction for diesel engines and as this paper will distinguish, has the ability to simplify direct injection fuel delivery systems in general.

This paper presents an empirical dynamic model of a single spring electromagnetic solenoid actuator system, including bounce, temperature effects and coil leakage inductance. The model neglects hysteresis and saturation, the aim being to compensate for these uncertainties through estimator robustness. The model is validated for all regions of operation and there is a good agreement between model and experimental data. A nonlinear (sliding mode) estimator is developed to estimate position and speed from current measurements. Since the estimator makes use of only current measurement it is given the name sensorless. The estimator is validated in simulation and experimentally. The novelty in this paper lies in the fact that accurate state estimation can be realized on a simple linear model using a robust observer theory. Also, the formulations for leakage inductance and coil temperature are unique.

This paper presents the modeling of an Electromagnetic Valve Actuator (EMV). A nonlinear model is formulated and presented that takes into account secondary nonlinearities like hysteresis, saturation, bounce and mutual inductance. The uniqueness of the model is contained in the method used in modeling hysteresis, saturation and mutual inductance. Theoretical and experimental methods for identifying parameters of the model are presented. The nonlinear model is experimentally validated. Simulation and experimental results are presented for an EMV designed and built in our laboratory. The experimental results show that sensorless estimation could be a possible solution for position control.

Advances in sensor solutions in the automotive sector make it possible to develop better ADAS and autonomous driving functions. One of the main tasks of highway chauffeur and highway pilot automated driving systems is to keep the vehicle between the lane lines while driving on a pre-defined route. This task can be achieved by using camera and/or GPS to localize the vehicle between the lane lines. However, both sensors have shortcomings in certain scenarios. While the camera does not work when there are no lane lines to be detected, an RTK GPS can localize the vehicle accurately. On the other hand, GPS requires at least 3 satellite connections to be able to localize the vehicle and more satellite connections and real-time over-the-air corrections for lane-level positioning accuracy. If GPS localization fails or is not accurate enough, lane line information from the camera can be used as a backup.

The Ohio State University has fielded teams at all three of the DARPA Grand Challenge and DARPA Urban Challenge autonomous vehicle competitions, using three very different vehicle platforms. In this paper we present our experiences in these competitions, comparing and contrasting the different requirements, strategies, tasks, and vehicles developed for each challenge. We will discuss vehicle control and actuation, sensors, sensor interpretation, planning, behavior, and control generation. We will also discuss lessons learned from the engineering and implementation process for these three vehicles.

The current Si/polymeric medical diagnostic sensors that are on the market only feature a one-point calibration system [1]. Such a measurement results in less accurate sensing and more in-factory sensor rejection. The two-point calibration fluidic method introduced here will alleviate some of the shortcomings of such current miniature analytical systems. Our fluidic platform is a disposable, multi-purpose micro analytical laboratory on a compact disc (CD) [2, 3]. This system is based on the centrifugal force, in which fluidic flow can be controlled by the spinning rate of the CD and thus a whole range of fluidic functions including valving, mixing, metering, splitting, and separation can be implemented. Furthermore, optical detection such as absorption and fluorescence can be incorporated into the CD control unit to obtain signals from pre-specified positions on the disc.

Over the last decade, many methods have been proposed for estimating the in-cylinder combustion pressure or the torque from instantaneous crankshaft speed measurements. However, such approaches are typically computationally expensive. In this paper, an entirely different approach is presented to allow the real-time estimation of the in-cylinder pressures based on crankshaft speed measurements. The technical implementation of the method will be presented, as well as extensive results obtained for a V-6 S.I. engine while varying spark timing, engine speed, engine load and EGR. The method allows to estimate the in-cylinder pressure with an average estimation error of the order of 1 to 2% of the peak pressure. It is very general in its formulation, is statistically robust in the presence of noise, and computationally inexpensive.

Autonomous vehicle technology has been developing rapidly in recent years. Vehicle parametric uncertainty in the vehicle model, variable time delays in the CAN bus based sensor and actuator command interfaces, changes in vehicle sped, sensitivity to external disturbances like side wind and changes in road friction coefficient are factors that affect autonomous driving systems like they have affected ADAS and active safety systems in the past. This paper presents a robust control architecture for automated driving systems for handling the abovementioned problems. A path tracking control system is chosen as the proof-of-concept demonstration application in this paper. A disturbance observer (DOB) is embedded within the steering to path error automated driving loop to handle uncertain parameters such as vehicle mass, vehicle velocities and road friction coefficient and to reject yaw moment disturbances.

This paper describes a systematic approach to identify the best sensor combination by performing sensor placement analysis to detect and isolate clutch stuck-off faults in Automatic Transmissions (AT) based on structural analysis. When an engaged clutch in the AT loses pressure during operation, it is classified as a clutch stuck-off fault. AT can enter in neutral state because of these faults; causing loss of power at wheels. Identifying the sensors to detect and isolate these faults is important in the early stage of the AT development. A universal approach to develop a structural model of an AT is presented based on the kinematic relationships of the planetary gear set elements. Sensor placement analysis is then performed to determine the sensor locations to detect and isolate the clutch stuck-off faults using speed sensors and clutch pressure sensors. The proposed approach is then applied to a 10-Speed AT to demonstrate its effectiveness.

This study is focused on exploring the possibilities of using camera and route planner images for autonomous driving in an end-to-mid learning fashion. The overall idea is to clone the humans’ driving behavior, in particular, their use of vision for ‘driving’ and map for ‘navigating’. The notion is that we humans use our vision to ‘drive’ and sometimes, we also use a map such as Google/Apple maps to find direction in order to ‘navigate’. We replicated this notion by using end-to-mid imitation learning. In particular, we imitated human driving behavior by using camera and route planner images for predicting the desired waypoints and by using a dedicated control to follow those predicted waypoints. Besides, this work also places emphasis on using minimal and cheaper sensors such as camera and basic map for autonomous driving rather than expensive sensors such Lidar or HD Maps as we humans do not use such sophisticated sensors for driving.

An important part of automotive driving assistance systems and autonomous vehicles is speed optimization and traffic flow adaptation. Vehicle sensors and wireless communication with surrounding vehicles and road infrastructure allow for predictive control strategies taking near-future road and traffic information into consideration to improve fuel economy. For the development of autonomous vehicle speed control algorithms, it is imperative that the controller can be evaluated under different realistic driving and traffic conditions. Evaluation in real-life traffic situations is difficult and experimental methods are necessary where similar driving conditions can be reproduced to compare different control strategies. A traditional approach for evaluating vehicle performance, for example fuel consumption, is to use predefined driving cycles including a speed profile the vehicle should follow.

A low cost system has been developed to measure a vehicle's lateral position relative to the lane markings on a roadway. The system is capable of tracking white or orange lines, solid or dashed edge lines, while operating in daylight or at night. The tracking system is comprised of two “off-the-shelf” black and white charge coupled device (CCD) video cameras along with commonly available electronic components. The lane tracking system is capable of outputting real time data at 30Hz through an analog output. Using the data from this sensor system it is possible to detect lane changes, determine the magnitude and duration of lane exceedances, and other metrics commonly used by researchers in the transportation community. This paper will discuss the design and performance of the system, processing of the raw lane tracker data, and the benefits and limitations of the technology.

The current approach for new Advanced Driver Assistance System (ADAS) and Connected and Automated Driving (CAD) function development involves a significant amount of public road testing which is inefficient due to the number miles that need to be driven for rare and extreme events to take place, thereby being very costly also, and unsafe as the rest of the road users become involuntary test subjects. A new development, evaluation and demonstration method for safe, efficient, and repeatable development, demonstration and evaluation of ADAS and CAD functions called Vehicle-in-Virtual –Environment (VVE) was recently introduced as a solution to this problem. The vehicle is operated in a large, empty, and flat area during VVE while its localization and perception sensor data is fed from the virtual environment with other traffic and rare and extreme events being generated as needed.

Automated driving has become a very promising research direction with many successful deployments and the potential to reduce car accidents caused by human error. Automated driving requires automated path planning and tracking with the ability to avoid collisions as its fundamental requirement. Thus, plenty of research has been performed to achieve safe and time efficient path planning and to develop reliable collision avoidance algorithms. This paper uses a data-driven approach to solve the abovementioned fundamental requirement. Consequently, the aim of this paper is to develop Deep Reinforcement Learning (DRL) training pipelines which train end-to-end automated driving agents by utilizing raw sensor data. The raw sensor data is obtained from the Carla autonomous vehicle simulation environment here. The proposed automated driving agent learns how to follow a pre-defined path with reasonable speed automatically.