Search Results

This work presents a method for estimating human body orientation using a combination of convolutional neural network (CNN) and stereo camera in real time. The approach uses the CNN model to predict certain human body keypoints then transforms these points into a 3D space using the stereo vision system to estimate the body orientations. The CNN module is trained to estimate the shoulders, the neck and the nose positions, detecting of three points is required to confirm human detection and provided enough data to translate the points into 3D space.

In practice, the piston wrist pin is either fixed to the connecting rod or floats between the connecting rod and the piston. The tribological behavior of fixed wrist pins have been studied by several researchers, however there have been few studies done on the floating wrist pin. A new bench rig has been designed and constructed to investigate the tribological behavior between floating pins and pin bore bearings. The experiments were run using both fixed pins and floating pins under the same working conditions. It was found that for fixed pins there was severe damage on the pin bore in a very short time (5 minutes) and material transfer occurs between the wrist pin and pin bore; however, for the floating pin, even after a long testing time (60 minutes) there was minimal surface damage on either the pin bore or wrist pin.

An existing challenge in current state-of-the-art autonomous vehicles is the process of safely transferring control from autonomous driving mode to manual mode because the driver may be distracted with secondary tasks. Such distractions may impair a driver’s situational awareness of the driving environment which will lead to fatal outcomes during a handover. Current state-of-the-art vehicles notify a user of an imminent handover via auditory, visual, and physical alerts but are unable to improve a driver’s situational awareness before a handover is executed. The overall goal of our research team is to address the challenge of providing a driver with relevant information to regain situational awareness of the driving task. In this paper, we introduce a novel approach to estimating a driver’s visual focus of attention using a 2D RGB camera as input to a Multi-Input Convolutional Neural Network with shared weights. The system was validated in a realistic driving scenario.

Computer simulation was used as a complement to crash and injury field data analysis and physical sled and barrier tests to investigate and predict the spinal injuries of a rear impact in an IndyCar. The model was expected to relate the spinal loads to the observed injuries, thereby predicting the probability and location of spinal fractures. The final goal is to help reduce the fracture risk by optimizing the seat and restraint system design and the driver's position using computer modeling and sled testing. MADYMO Full FE Human Body Model (HBM) was selected for use because of its full spinal structural details and its compatibility with the vehicle and restraint system models. However, the IndyCar application imposed unique challenges to the HBM. First, the driver position in a race car is very different from that in a typical passenger car.

Production software validation is critical during software development, allowing potential quality issues that could occur in the field to be minimized. By developing automated and repeatable software test methods, test cases can be created to validate targeted areas of the control software for confirmation of the expected results from software release to release. This is especially important when algorithm/software development timing is aggressive and the management of development activities in a global work environment requires high quality, and timely test results. This paper presents a hardware-in-the-loop (HIL) test bench for the validation of production transmission controls software. The powertrain model used within the HIL consists of an engine model and a detailed automatic transmission dynamics model. The model runs in an OPAL-RT TestDrive based HIL system.

During the production controller and software development process, one critical step is the controller and software verification. There are various ways to perform this verification. One of the commonly used methods is to utilize an HIL (hardware-in-the-loop) test bench to emulate powertrain hardware for development and validation of powertrain controllers and software. A key piece of an HIL bench is the plant dynamics model used to emulate the external environment of a modern controller, such as engine (ECM), transmission (TCM) or powertrain controller (PCM), so that the algorithms and their software implementation can be exercised to confirm the desired results. This paper presents a 6-speed automatic transmission plant dynamics model development for hardware-in-the-loop (HIL) test bench for the validation of production transmission controls software. The modeling method, model validation, and application in an HIL test environment are described in details.

This paper presents a hardware-in-the-loop (HIL) test bench for the validation of production transmission controls software, with a focus on a closed-loop vehicle drive-train model incorporating a detailed automatic transmission plant dynamics model developed for certain applications. Specifically, this paper presents the closed-loop integration of a 6-speed automatic transmission model developed for our HIL transmission controller and algorithm test bench (Opal-RT TestDrive based). The model validation, integration and its application in an HIL test environment are described in details.

Application of cervical spine range of motion data and related anthropometric measures of the head and neck include physical therapy, product design, and computational modeling. This study utilized the Cervical Range of Motion device (CROM) to define the normal range of motion of the cervical spine for subjects five (5) through ten (10) years of age. And, the data was collected and analyzed with respect to anatomical measures such as head circumference, face height, neck length, and neck circumference. This study correlates these static anthropometric measures to the kinematic measurement of head flexion, extension, lateral extension, and rotation.

This paper provides an overview of rollover crash safety, including field crash statistics, pre- and rollover dynamics, test procedures and dummy responses as well as a bibliography of pertinent literature. Based on the 2001 Traffic Safety Facts published by NHTSA, rollovers account for 10.5% of the first harmful events in fatal crashes; but, 19.5% of vehicles in fatal crashes had a rollover in the impact sequence. Based on an analysis of the 1993-2001 NASS for non-ejected occupants, 10.5% of occupants are exposed to rollovers, but these occupants experience a high proportion of AIS 3-6 injury (16.1% for belted and 23.9% for unbelted occupants). The head and thorax are the most seriously injured body regions in rollovers. This paper also describes a research program aimed at defining rollover sensing requirements to activate belt pretensioners, roof-rail airbags and convertible pop-up rollbars.

Automotive rollover is a complex mechanical phenomenon. In order to understand the mechanism of rollover and develop any potential countermeasures for occupant protection, efficient and repeatable laboratory tests are necessary. However, these tests are not well understood and are still an active area of research interest. It is not always easy or intuitive to estimate the necessary initial and boundary conditions for such tests to assure repeatability. This task can be even more challenging when rollover is a second or third event (e.g. frontal impact followed by a rollover). In addition, often vehicle and occupant kinematics need to be estimated a-priori, first for the safe operation of the crew and equipment safety, and second for capturing and recording the event. It is important to achieve the required vehicle kinematics in an efficient manner and thus reduce repetitive tests. Mathematical modeling of the phenomenon can greatly assist in understanding such kinematics.

Implementation of engine turnoff at idle is desirable to gain improvements in vehicle fuel economy. There are a number of alternatives for implementation of the restarting function, including the existing cranking motor, a 12V or 36V belt-starter, a crankshaft integrated-starter-generator (ISG), and other, more complex hybrid powertrain architectures. Of these options, the 12V belt-alternator-starter (BAS) offers strong potential for fast, quiet starting at a lower system cost and complexity than higher-power 36V alternatives. Two challenges are 1) the need to accelerate a large engine to idle speed quickly, and 2) dynamic torque control during the start for smoothness. In the absence of a higher power electrical machine to accomplish these tasks, combustion-assisted starting has been studied as a potential method of aiding a 12V accessory drive belt-alternator-starter in the starting process on larger engines.

In 2002, NHTSA statistics indicate air bag deployments saved an estimated 1,500 lives; however, reports of occupants having serious or fatal injuries during air bag deployment appear low relative to the number of accidents with air bag deployments. To avoid air bag induced injuries, a variety of occupant sensing technologies are being developed. One of the critical logic deployment challenges faced by these technologies is whether the system can accurately determine if the occupant is in a posture or a position such that air bag deployment may result in an injury. To improve accuracy, it is necessary to understand what postures the occupants are likely to assume during a ride and how often. For this purpose, Delphi Corporation has conducted a survey to solicit opinions on the posture usage rate. With 560 responses, the frequencies for 29 sitting postures for adult passengers and 13 child postures or positions were estimated.

This paper presents the development of coordinated control of vehicle systems, specifically for controlled brakes and controlled steering systems. By utilizing a control structure to oversee a four-wheel-steer (4WS) system and a brake-based vehicle stability enhancement (VSE) system, it is possible to achieve improvements in vehicle stability and driver workload/comfort, and to reduce compromises in vehicle handling. The coordinated control is designed to leverage the unique strengths of 4WS and VSE, and to prevent conflicts between them. Vehicle test results prove the viability of the concept.

A RTD (resistive temperature device) high temperature sensor was developed for exhaust gas temperature measurement. Extensive modeling and optimization was used to supplement testing in development. The sensor was developed to be capable of withstanding harsh environments (-40° to 1000°C), typical of engine applications, including poisons, while maintaining high accuracy (< 0.5% drift after 500 hrs of aging at 950°C). The following sensor characteristics are presented: resistance-temperature curve, accuracy, response time, and long-term durability. In addition, a system error analysis program was developed with representative results.

The industry strategy for automotive safety systems has been evolving over the last 20 years. Systems, such as frontal and side airbags, are available today on the worldwide market that provide proven safety benefits. Interest in advanced safety systems for occupant protection and accident avoidance, is focused on making further reductions in road fatalities and injuries. Interior occupant sensing systems for advanced restraint systems, trapped occupant sensing, and driver monitoring are today under intense development as part of the industry's safety vision. In this paper, we will discuss the need for and requirements of interior occupant sensing systems, as well as applicable technologies

A seat belt usability study was conducted to investigate factors associated with seat belt comfort and convenience related to shoulder belt contact pressure, shoulder belt fit, and seat belt upper anchorage location. Two major objectives were addressed in this study: (1) Determine the shift in the contact pressure while changing the seat back angle and seat belt attachment points / B-pillar location by utilizing a body pressure measurement system; (2) Identify how seat belt contact pressure and fit affect users' subjective feeling of comfort. Results from the statistical analysis shows that the seat belt contact pressure increases when the D-ring moves away from the driver in the fore-aft direction (X-axis) whereas height adjustment of the D-ring (Z-axis) is not statistically significant in terms of pressure distribution.

As an essential dimension of product harmony and craftsmanship, product sound quality has drawn increasing attention from customers in recent years. To meet this customer requirement, Delphi Corporation has been taking a proactive role in understating customer preferences, improving designs, and developing a sound quality knowledge base for this purpose. This study investigates the characteristics of the glovebox closure sound that affects the customer's perception of the product harmony and craftsmanship. Previous research has indicated that the perceived closure sound quality is affected by the spectral balance, the occurrence of multiple impulses, and the duration of the closing event. The primary goal of this study is to explore how these parameters affect the perception of glovebox sounds and to what extent. A jury evaluation was conducted with a sequence of glovebox closure sounds, which were derived from an existing recording.

Electro-hydraulic actuation has been used widely in automatic transmission designs. With greater demand for premium shift quality of automatic transmissions, higher pressure control accuracy of the transmission electro-hydraulic control system has become one of the main factors for meeting this growing demand. This demand has been the driving force for the development of closed loop pressure controls technology. This paper presents the further research done based upon a previously developed closed loop system. The focus for this research is on the system requirements, such as solenoid driver selection and system latency handling. Both spin-stand and test vehicle setups are discussed in detail. Test results for various configurations are given.

Proper valve angles and concentric valve seats are critical to performance of an engine. If the valve seat were not right, the valve is not going to seat properly resulting in reduced power output. Although the performance of CNC machines is accurate, unavoidable human errors at the part loading position have serious repercussions on engine performance. A solution algorithm presented in this paper employs the principles of inverse kinematics wherein a faulty compound-hole angle axis in space caused by the translational and rotational errors at the part loading position is identified with an imaginary true axis in space by enforcing identity through a modified machine axes.

In occupant sensing system development, the Anthropomorphic Test Dummy (ATD) and the Occupant Classification ATD (OCATD) are frequently used to simulate live human subjects in the testing and validation of weight based occupant sensing systems. A study was conducted to investigate the range of loading differences between these ATDs and live human subjects over various seating postures and conditions. The results of the study revealed that differences in seat load patterns could be significant, even though both the ATD and live humans are in the same weight and body size categories. Seat loading was measured using Hybrid III (5th percentile female, 50th percentile male, and 3 year old) ATDs, OCATDs (OCATD5 - 5th percentile female, and OCATD6 - 6 yr old child), and a CRABI (12-month old) dummy. Human subjects in the same weight and height categories as the above listed ATDs were also measured.