The Forward-looking Laser Sensor (FLLS) was developed to explore the benefits of laser technology as an alternative to millimeter wave radar for automotive headway sensors. It was felt that a laser based approach might be less costly and could potentially offer performance advantages. Headway sensors are a key element for future collision avoidance and intelligent cruise control systems. The specific task objectives were:
The system described here (Engineering Development Unit Version 1 or EDU1) is for development only. As such, it provides many development features to support the definition of a production intent system. A vendor with experience in low-cost pulsed laser rangefinders carried out most of the sensor development.
A raster scanned laser rangefinder was designed, built, and tested in the laboratory and on a vehicle. The range to 375 points in a unique interlaced raster (as illustrated in Figure 3.36) are determined every 150 ms. The wide (20X8) field of view allows multiple targets in 3 lanes to be identified and tracked in both azimuth and elevation within the sensor's 100 meter range. The ability to overcome some of the problems found in existing laser rangefinder systems for collision avoidance was demonstrated.
Figure 3.36: Raster Scanned Laser Sensor
Raster scanned laser rangefinders may ultimately provide the high performance and low cost necessary for automotive collision avoidance applications. The engineering development sensor reported here is a first step toward designing a production sensor. The sensor uses pulsed time-of-flight laser rangefinder technology to show the feasibility of this approach. In addition, the development sensor includes data capture features to evaluate various levels of scanning so that the most cost-effective approach can be selected. Several diagnostic capabilities are included to support development. A video camera records real time traffic events with a sensor data overlay and a PCMCIA memory card interface is available for high-speed sensor data capture.
The basic requirements for a forward-looking laser sensor are as follows:
1. The system must simplify the driving process.
2. The system must react correctly to all targets found on public and private roads that could pose a threat to the safe operation of the vehicle.
3. The system must enhance the safety of driver and passengers.
Using these requirements as guidelines, a scanning laser rangefinder based on time-of-flight rangefinder technology was selected. A narrow (20 Ns) pulse of light from a semiconductor laser is timed as it travels from the transmitter to the target and returns. This is a relatively mature technology with thousands of military and commercial systems in service. The basic criteria for the design selection were: (1) the ability to detect non-cooperative automotive targets at 100 meters, and (2) .the system be eye safe.
The FLLS system uses the time-of-flight rangefinding technique. The distance to the target is obtained by measuring the time interval between the transmitted and received light pulses. To measure this time interval a high-speed counter is started by a trigger pulse when the laser diode is fired. A signal generated by the receiver when a light pulse is received is used as an enable signal to latch the current value of the counter into a location in a FIFO (first in first out) memory device. After the field of view has been scanned, the contents of the FIFO are then analyzed to determine the ranges of returns based on the counter values loaded into the FIFO. These counter values are a record of the total time of flight and thus the range to the targets. This analysis is done with assembly language subroutines highly optimized to take full advantage of the more powerful special features of the TI C31 processor. This raw range data is then transmitted to the target tracking PC for further analysis as a sequence of 25 range values each corresponding to a different azimuth angle in the single line sweep.
During vehicle testing phase, a software was developed and used to create and track targets. The system has demonstrated the ability to create targets from pixels and track multiple targets in 3 lanes of traffic. Depending on the traffic, the system has been observed to track up to 4 and 5 targets simultaneously. These traffic scenarios were captured on videotape using a vehicle system depicted in Figure 3.37.
Figure 3.37: FLLS Vehicle System Mechanization
Back -- TOC -- Forward