Briefings were prepared and presented on the Prototype Vehicle Validation Plan and discussions were held with NHTSA, Volpe, and other government representatives, on system level testing scenarios. A Test Scenarios Report was delivered to NHTSA in April 2000. A System Verification Plan will be delivered to NHTSA in December 2000.

Each subsystem was reviewed to develop a preliminary hazard analysis and create a hazard mitigation plan. The plan is being developed using guidelines from Mil Standard 882C and SAE J1789. Safety plan presentations were made at meetings in November 1999 and June 2000. The Risk Management Plan will be delivered to NHTSA in January 2001.

The Forward Radar Sensor Task includes transceiver/antenna integration, and development of algorithms for auto-alignment, antenna radome blockage detection, and bridge rejection. To facilitate transceiver/antenna integration, millimeter-wave monolithic integrated circuits (MMIC) were designed into the transceiver. Initial tests of the MMIC components and of the entire radar on an engineering development vehicle found some performance issues that are being addressed. The unit for the Prototype Vehicle is expected to perform as specified.

The auto-alignment algorithm is to electronically adjust the sensor mechanisms for misalignment due to vehicle wear and tire alignment. The basic algorithm has been found effective over "long" periods of time but is susceptible to peak errors of short duration when the vehicle is driven on curves. Engineering testing is continuing to isolate the cause of these errors and to develop a remedy.

The radome blockage detection algorithm is to detect when dirt, slush, or other material blocks the sensor. Development of this algorithm is behind schedule but a recovery plan is being executed. Some conceptual concerns have been identified regarding detection reliability in partial blockage situations, heavy snowfall, or when the vehicle is parked. These will be investigated and appropriate validation tests will be defined and executed.

The bridge rejection algorithm is to recognize and classify bridges as safe overhead objects so they do not cause the ACC function to slow the vehicle. Several approaches have been tested. An amplitude-slope method was found to be reliable but it requires multiple scans with range closure that can lead to delayed recognition of valid in-path stopped vehicles.

The forward vision sensor is to provide lane tracking to help distinguish in-path from out-of-path targets, which is particularly difficult as the vehicle approaches a curve and during lane changes. It will use a video camera mounted behind the windshield of the vehicle to estimate road shape, lane width, vehicle heading and lateral position in the lane. Teams from the University of Pennsylvania, Ohio State University, and the University of Michigan – Dearborn were contracted to enhance their existing technology to meet the needs of the FOT. The primary challenge is to provide adequate road shape estimates at least 75 meters (preferably 100 meters) ahead of the vehicle. The work of the three universities will be evaluated to select one approach for further development and final integration into the FOT vehicles.

During the first year, Delphi-Delco Electronics defined requirements, implemented a video data acquisition system, and worked with the universities to define appropriate confidence measures. The universities have demonstrated systems that can track highway roads at 10 Hz frame rates out to 75 meters. They can handle lane changes and partial occlusion of the lane markings. Work is still required to improve tradeoffs associated with sampling schemes and to improve performance of the lane marker extraction methods with low levels of illumination and on concrete road surfaces.

A new Delphi Brake Control System will replace the OEM brake components on the Prototype and FOT deployment vehicles. The brake control system includes an anti-lock brake system (ABS), vehicle stability enhancement, and traction control features. For this program, the brake system will be enhanced to respond to ACC braking commands while maintaining the braking features and functions that were in the original brake system. Delphi's common best engineering practices will be used to perform safety analysis and vehicle level verification of the brake system to ensure production-level confidence in the brake system.

Over the past two years, the DBC 7.2 brake control system has undergone significant testing for production programs. During the first year of the ACAS/FOT program, the brake system was integrated on a chassis mule and one of the Engineering Development Vehicles. Calibration and tuning of the brake system has started.

The throttle control system maintains the vehicle speed in response to the speed set by the driver or in response to the speed requested by the ACC function. The Delphi stepper motor cruise control (SMCC), standard in the Buick LeSabre, will be modified to perform the required functions. The required modifications have been used successfully in other projects. During the first year, interface requirements were defined and throttle control system modifications were designed for the prototype vehicle.

The driver vehicle interface senses the settings from the driver via buttons on the steering wheel. It also conveys information from the ACC and FCW functions to the driver. The FCW warnings must immediately direct the driver to evaluate and react to threats with sufficient time to perform some action to avoid or mitigate a potential crash. To achieve this, audible, visible, and possibly haptic cues will be employed. The ACC information must be presented so that the driver can easily determine the cruise control set speed, selected inter-vehicle separation, and whether or not a preceding vehicle has been detected by the system. For both functions, this information must be understandable at a glance by the driver and without adding extra workload to the driving task

The primary visual interface for the FOT system will be a reconfigurable, high-resolution, full-color head-up display. During the first year of the program development of hardware began, including an agreement with a manufacturer for the visual display cells. To support development, a Buick LeSabre was procured as a test bench.

Candidate display formats were developed for three warning philosophies: a single-stage imminent alert, a two-stage warning, and a graded display with an imminent alert. The alternatives differ in the amount of information provided to the driver at times other than when immediate action is required. The differences may impact whether drivers are startled or annoyed by the warnings, and their vehicle following behavior. The graded display philosophy is the preferred alternative for the ACAS/FOT program, though interfaces for the other two approaches will be designed and evaluated in driving simulator and test track scenarios.

The warning philosophy alternatives will be refined by testing them using subjects in driving simulators, on test-tracks and public roads. A driving simulator at Delco was upgraded to support refinement of the visual aspects of the displays. Driving simulator tests conducted at the University of Iowa and on-road experiments at GM's Milford Proving Ground will also be used to provide empirical data to help refine warning timing and to select the final warning philosophy. Development of the test protocols has begun in collaboration with NHTSA. A DVI Warning Cue Implementation Summary Report will be submitted to NHTSA in February 2001.

Data fusion algorithms will be used to provide estimates for road geometry using several sources of information. These sources of information include the vision sensor, scene tracking, yaw rate and speed sensors, and map-based road geometry estimation. Scene tracking is a technique to estimate road geometry by tracking the path of vehicles detected by the forward-radar sensor. Map-based road geometry estimates will be derived from data extracted from digital maps using GPS and dead reckoning to derive the vehicle's geographic location. Data fusion techniques will also be used to evaluate sensor information to modify the expected driver reaction time and braking intensity based upon the environmental conditions (rain, snow, day, night, etc.) and/or driver activity such as adjusting the climate control system.

During the first year of the program, requirements were developed and appropriate data fusion algorithmic approaches were selected. A new road model parameterization technique using multiple clothoids was developed and found to provide smaller errors than single-clothoid road models, particularly near transition between straight and curved road segments. Better road geometry estimates should translate into lower errors in distinguishing in-path targets from out-of-path targets.

Target identification and selection uses the road geometry estimate to determine which radar returns are from objects that are (or will soon be) in the path of the vehicle. The selected targets are evaluated by the FCW function to decide whether to issue an alert. The Tracking and Identification Task includes development of the algorithms for scene tracking, map-based road geometry estimation, path estimation, target identification and selection.

In the first year of the program, the target identification algorithms were enhanced with improvements to the path prediction algorithm, lane change detection, and roadside distributed stopped object detection. These were tested with an improved tool for simulation of road scenarios. Scene tracking algorithm work included enhancement of the algorithms that handle vehicles that are not following the lanes, real-time implementation, and tuning with real-world radar target data. Map-based road geometry estimation work included completion of the sensor driver software and map data retrieval software. Limited testing of this software was completed with promising results. Assistware delivered a map enhancement approach in February 2000. This technology tracks the vehicle's route using a GPS receiver to augment the road geometry information in the map database.

To support development of the hardware and software Delphi Delco Electronics created three engineering development vehicles (EDV), by modifying them to provide the functionality of a rudimentary integrated ACC and FCW system.

The CW Function Task includes In-Vehicle Threat Assessment Algorithm Development and Threat Assessment Simulation. The threat assessment algorithm assigns a threat level to the current situation based upon the motion of the project vehicle and each selected in-path target vehicle. Six threat assessment algorithms, including one developed by NHTSA, will be implemented and tested in simulations, on test tracks, and in real traffic. Work in the first year of the program included development and analytical evaluation of the algorithms.

The Threat Assessment Simulation is for refinement and evaluation of threat assessment algorithms. The University of California-PATH, using mathematical models of each function provided by GM, is developing it. Initial coding of the simulation was completed in August 2000.

The Adaptive Cruise Control Subsystem is a module from a future GM production program that includes the radar and ACC Controller. The Adaptive Cruise Control Function Task will provide the module as is for the 2002 Buick LeSabre and provide support for its integration with the rest of the ACC and FCW functions. The work is focussed on the interfaces between the module and other vehicle subsystems.

During Phase I of the program the Fleet Vehicle Build Task includes building and testing a GM Engineering Development Vehicle (GM EDV) and a Prototype Vehicle. The GM EDV is a 2000 Buick LeSabre with modifications to accommodate all the required instrumentation to investigate threat assessment, map-based path prediction, map database enhancement, and human factors. During the first year of the program the GM EDV vehicle was modified to incorporate required changes to the electrical and mechanical infrastructure, and software was developed for the interfaces. The system hardware was debugged and installed in the vehicle after which operation of the sensors was verified.

The Prototype Vehicle will integrate all the technologies developed by the partners in the program as a precursor to the Pilot Vehicles and finally the Deployment Vehicles. The Prototype will have the functionality, but not necessarily the form factor, of the deployment vehicles. It will be used to verify the functionality of the ACC and FCW features during Phase I of the program. The Pilot Vehicles will have the functionality and form factor of the deployment vehicles and will be built during Phase II of the program. All materials to be installed in the Prototype Vehicle have been ordered and an early version of the ACC brake system has been installed. Installation of the remaining equipment will begin shortly and will require significant collaboration among the partners to complete.

The Field Operational Test (FOT) task includes preparation and execution of the test itself. In Phase I of the program this task includes planning and performing two stages of pilot tests, development of a Data Acquisition System (DAS) and developing the procedures, software, and a plan for execution of the FOT. Much of this work is being performed by University of Michigan Transportation Research Institute (UMTRI) staff, who will apply prior methodology and learning from the Intelligent Cruise Control (ICC) FOT, adapting the field-testing techniques to the ACAS platform.

In June of 2000, eight UMTRI staff with prior experience testing ACC systems evaluated one of the Delco EDVs over a 94-mile route. The ACC system was found to be highly operable and was successfully driven with the ACC engaged for approximately 90% of the mixed-route miles and for over 80% of the miles on surface streets. The intent was to explore and identify challenging conflict types that occur in the roadway environment.

Also during the first year of the program a Data Acquisition System that includes many, but not all, of the features and software of the eventual FOT package was successfully constructed and operated on the Delco EDV. The system collected many of the variables required for the FOT including the multi-target radar data. This operation confirmed the readiness of the DAS for handling the tasks of data collection in the FOT.

Table 1.2 shows the major milestones that were accomplished during the first year of the program. Table 1.3 shows the milestones for the remainder of the year.