Search Results

For the autonomous driving and the ADAS, we have been developing the new vision system. It focuses on detecting the small object at far ranges and keeps the vehicle running smoothly by avoiding the object in advance. This system is based on the high-resolution monocular camera with narrow FOV and the core algorithms for object-detection and lane-detection. Since we have already developed the prototype system, here we have applied it to a field test. The test results are: for traffic cones and arbitrary objects at 50 - 150 [m], the detection rate is 95.6% for n=90, the ghost rate is 0% for n=53, and the range accuracy is 7.1 [m] for n=72. To attain this performance, we also have developed the support algorithms/methods for deleting a ghost object caused by a puddle, determining the side line for an object with obscure boundary and estimating the elevation angle.

For one of the advanced sensing technologies for the autonomous driving, we have been working on the new vision system. It focuses on detecting the small object at far ranges. It enables to detour a vehicle by avoiding small object. This system is based on the high-resolution mono-camera with narrow FOV and the algorithms for object-detection and lane-detection. Since we have already proposed the system for the traffic cones, we proceed to a new algorithm for an unknown object with arbitrary shape. It can detect the object and estimate its range by using the information on the lane even though the shape and dimension of the object are unknown. Here the system architecture and how the algorithm to detect the object and estimate its range are described. The algorithm is at first validated by a computer-generated image. And then it is applied to real images. For the cones, the accuracy was +/- 4 [m] for the ranges between 50 and 150 [m].

As one of the advanced sensing technologies of Autonomous driving, we have started developing the new vision system. It focuses on detecting the small object at far ranges. It makes it possible to detour a vehicle along the traffic cones (small object). This system is based on the high-resolution mono-camera with narrow FOV and the algorithms for object-detection/ranging and lane-detection. A prototype is created and evaluated. It could detect the cones and estimate their ranges up to 150 [m]. The range accuracy is 5.3 [m]. Plural cones at different ranges can be measure at one time. The lane detection could reach 150 [m].

A practical and low cost Blind Spot Monitoring system is proposed. By using a single camera, the range and azimuth position of a vehicle in a blind spot are measured. The algorithm is based on the proposed RWA (Range Window Algorithm). The camera is installed on the door mirror and monitoring the side and rear of the host vehicle. The algorithm processes the image and identifies range and azimuth angle of the vehicle in the adjacent lane. This algorithm is applied to real situations. The 388 images including several kinds of vehicles are analyzed. The detection rate is 86% and the range accuracy is 1.6[m]. The maximum detection range is about 30[m].

A method, which determines the range and lateral position of a preceding vehicle on the road by a single image, had been proposed. It is based on the Range-Window algorithm (RWA) and Pattern Matching (PM). The RWA estimates the range by using multiple virtual windows of fixed physical size at different distances. The size ratio between the windows and a preceding vehicle determines the best window. The associated range of the window will be the range of the vehicle. The PM complementarily estimates the range by using a template obtained through the RWA. It works especially well when an occlusion occurs due to shadows of road side objects. The range estimation is based on the use of horizontally-modified patterns of the template. Namely this PM can perform the range estimation as well as the object extraction. Here, for the real time operation, the calculation time of this method was evaluated after coding the RWA and PM into C-language (16 bit calculation) from Matlab (64 bit).

An algorithm, which determines the range of a preceding vehicle by a single image, had been proposed. It uses a “Range-Window Algorithm”. Here in order to realize higher robustness and stability, the pattern matching is incorporated into the algorithm. A single camera system using this algorithm has an advantage over the high cost of stereo cameras, millimeter wave radar and non-robust mechanical scanning in some laser radars. And it also provides lateral position of the vehicle. The algorithm uses several portions of a captured image, namely windows. Each window is corresponding to a predetermined range and has the fixed physical width and height. In each window, the size and position of objects in the image are estimated through the ratio between the widths of the objects and the window, and a score is given to each object. The object having the highest score is determined as the best object. The range of the window corresponding to the best object becomes an estimated range.

Principle of the target tracking method for the Adaptive Cruise Control (ACC) system, which is applicable to non-uniform or transient condition, had been proposed by one of the authors. This method does not need any other information rather than that from the radar and host vehicle. Here the method is modified to meet more complex traffic scenarios and then applied to data measured on real highway. The modified method is based on the phase chart between the lateral component of the relative velocity and azimuth of a preceding vehicle. From the trace on the chart, the behavior of a preceding vehicle is judged and the discrimination between the lane change and curve-entry/exit can be made. The method can deal with the lane-change of a preceding vehicle on the curve as well as on the straight lane. And it is applied to more than 20 data including several road/vehicle conditions: road is straight, or turns right or left; vehicles are motorbikes, sedans and trucks.

An algorithm, which determines the range of a vehicle on the road by a single image, is proposed. Since it uses a single camera, there is not a problem like the high cost in the stereo camera, the millimeter wave radar and the non-robust mechanical scanning in the laser radar. And it also gives the lateral position of the vehicle. So far it is difficult to get the range information from a single camera. The algorithm overcomes this drawback by introducing a new concept, “Range Window”. It uses several portions of a captured image, namely windows. Each window is corresponding to a certain pre-determined range and has the fixed physical width and height. In each window, the size and position of an object in the image is estimated through the ratio between the widths of the object and the window and a score is given to the window. The window having the best score is determined as a best window. The range corresponding to the best window becomes an estimated range.

A new practical algorithm is proposed for multiple object detection in automotive FM-CW radars. They are radars for ACC (Adaptive Cruise Control) radar, collision avoidance, pre-crash safety, side-object detection, etc. This algorithm can provide the distance and relative velocity of objects without the ambiguity of distance and relative velocity, an inherent problem of FM-CW. Since it is simple, straight-forward and fast, it is suitable for automotive application, in which the update time is less than 100 [msec]. This algorithm is based on two down chirp frequency sweeps with small slope-difference. Since the difference is small, the correct pairs of beat frequencies are obtained automatically. Because of the down chirps, the polarity of beat frequencies owing to the distance and the doppler becomes the same for an approaching object and then the distance and the velocity are uniquely determined.

In the Adaptive Cruise Control (ACC) system, the conventional target tracking method depends on the yaw rate/steering angle information and is effective for the uniform road condition, where the host and target vehicles are on the roads with the same curvature. But it cannot work on non-uniform road conditions where the target vehicle and host vehicle are on the road with different curvature like curve-entry and lane change. Therefore a new method to track a target vehicle in non-uniform conditions has been developed. It is based on the locus on the phase chart between the azimuth angle and relative velocity, which are obtained from the radar and host vehicle information. It can express the path of the target vehicle in non-uniform conditions exactly since the locus can be expressed theoretically. The theory of the method and the simulation results are shown here. The performance for the curve-entry, curve-exit and lane-change is satisfactory.

A new method to simulate “the FM radio reception on the road” in an open space is proposed. It can provide reliable and objective evaluation of the radio system including a vehicle. It can reduce the cost and time of the driving test on roads. It is based on two technologies. One is a wave-extraction system of incoming waves on the roads and the other the field reconstruction in an open space. For the extraction, 3-D software based on AIC is developed and can estimate the number of waves and the amplitude, phase and polarization of the waves from the field data. The software showed the extraction accuracy of 90% in simulation. This method was applied to the road test, and it was found that the software could extract the waves. And through several tests in the space, the feasibility of the reconstruction of standing wave was confirmed (Maximum-minimum ratio: 18 dB and 22 dB at FM-band for the horizontal and the vertical polarization).

A new type of directional antenna, to be installed in a vehicle, is proposed to achieve an acceptable gain, fading reduction and the suppression of signal distortion for the FM broadcasting reception. This antenna consists of several antenna elements, a phase-synthesis circuit and a switching system. The circuit synthesizes the four patterns which are oriented toward the front, rear, left and right directions of the vehicle, respectively. The system switches the patterns from one direction to another, depending on the received voltage at a time. A prototype directional antenna, with four elements installed on the front and rear glasses of a vehicle, was fabricated and field-tested. This antenna had the smaller fading than the Pole one by 7 dB and the same fading with the diversity. In the weak field the lowest level, the 1% point of the distribution of the received voltage, was better in the directional one than in the Pole one by 3 dB.

A new type of antenna, installed on the rear glass of a vehicle, to receive both Band III and L band was developed. This antenna consists of a rectangular loop, inductance elements inserted in the loop and a monopole connected to the edge of the loop. It works, through the use of the elements, both as a one-wavelength loop antenna for Band III and a half-wave dipole for L band. This antenna is (1) conformal and (2) compatible with the defogger heater grid lines. As the results of an open site test, this antenna had a higher gain than that of commercially- available whip-type multiband antennas for both bands.