Fabian Roos

Reliable Orientation Estimation of Vehicles in High-Resolution Radar Images

With new generations of high-resolution imaging radars, the orientation of vehicles can be estimated without temporal filtering. This enables time-critical systems to respond even faster. Based on a large data set, this paper compares three generic algorithms for the orientation estimation of a vehicle. An experimental MIMO imaging radar is used to highlight the requirements of a robust algorithm. The well-known orientated bounding box and the so-called L-fit are adapted for radar measurements and compared with a brute-force approach. A quality function selects the best fitted model and is a key factor to minimize alignment errors. Moreover, the reliability of the estimation is evaluated with respect to the aspect angle, the distance to the target, and the number of sensors. An approach to estimate the reliability of the current orientation estimation is introduced. It is shown that the root mean square error of the orientation estimation is 9.77° and 38% smaller compared with the common algorithm. In 50% of the evaluated measurements the orientation estimation error is smaller than 3.73°.

Adaptive clustering for contour estimation of vehicles for high-resolution radar

Future automotive radars will be able to achieve much higher range and angular resolution compared to currently used radar sensors. This enables functionalities like vehicle contour estimation to be used in advanced driver assistance systems, thus heavily increasing their performance. In this paper, the application of an adaptive algorithm on basis of k-nearest-neighbours examination for clustering radar data as precursor to estimation of width, length, and position of vehicles is presented and compared to a more basic algorithm. The influence of the parameters of this KNN-DBSCAN algorithm and its performance dependency on the used MIMO radar system is discussed.

Stacking PS-QPSK and 64PPM for Long-Range Free-Space Transmission

In this paper a new sensitivity record for uncoded transmission with 2.6 photons per bit (4.15 dB) is presented. This is achieved using four dimensional (4-D) stacked orthogonal modulation formats i.e. PS-QPSK-64PPM.

Range migration compensation for chirp-sequence based radar

To improve the range and Doppler resolution the bandwidth and the observation time need to be increased. This leads to a range migration effect and thus to a declining separability of targets. This paper compares three different modulation formats to cope with the range migration. The standard chirp-sequence modulation, the bandwidth variation modulation known from literature, and the proposed chirp duration variation are analysed. A simulation is used to validate the modulation format.

Enhancement of Doppler resolution for chirp-sequence modulated radars

For automotive radar sensors the Doppler resolution is a key parameter, since it is used to separate targets which are in the same radial distance. For future applications the Doppler signature of targets can be exploited for classification purpose, e.g. discrimination among road users like pedestrians or bicyclists. In this paper, a signal processing scheme based on the chirp-sequence modulation principle is proposed to enhance the Doppler domain resolution. Simulations and measurement results are shown to prove the signal processing leading to an enhanced separability of close targets.

Waveform multiplexing using chirp rate diversity for chirp-sequence based MIMO radar systems

MIMO radar systems create a large virtual aperture to enhance the angular resolution. As a multiplexing scheme often the time-division multiplexing (TDM) procedure is chosen. The drawbacks are a reduced maximal unambiguously detectable Doppler frequency and the need to correct a phase error in angle estimation for relative radial velocities. To overcome these disadvantages a multiplexing scheme is proposed which uses different chirp rates. Every transmitting antenna is active at the same time but can be distinguished due to the different slopes of the frequency ramps.

Blind Adaptive Beamforming for Automotive Radar Interference Suppression

Radar sensors offer enormous advantages as sensing devices for automated and autonomous driving. However, when multiple of such sensors are operated in a large number of cars, there is a high risk of the occurrence of mutual interference. In the currently widespread linearly frequency modulated sensors interference reduces the detection performance, especially for targets with a low radar cross section. In this paper the interference effects are suppressed with an adaptive beamforming scheme based on a mean square error minimization. The paper evaluates the algorithm with the help of a simulated and measured scenario, and discusses the occurring interference effects and the benefits of the beamforming approach.

Data rate reduction for chirp-sequence based automotive radars using compressed sensing

For autonomous driving high-resolution radar sensors are key components, which have the drawback of high data rates. In order to reduce the amount of sampled data, random samples can be omitted and afterwards reconstructed using compressed sensing methods. A possible application is that not every receiving antenna element demands its own analog-to-digital converter. One converter can be used for several receiving elements with a random assignment to each antenna instead. In this paper, an analysis is presented of how many samples can be neglected such that a successful reconstruction in post-processing for an automotive scenario is possible. A measurement result is shown to prove that with only 40 % of samples a successful reconstruction is possible.

Ghost target identification by analysis of the Doppler distribution in automotive scenarios

In an automotive environment the presence of reflecting surfaces cannot be avoided. The electromagnetic wave returning from a target vehicle can get reflected on those surfaces causing a non existing so-called ghost target. For driver assistance systems ghost targets can lead to false decisions and, therefore, they should be detected and avoided. In this paper a model for describing those ghost targets and a procedure to distinguish them from real targets using the orientation and the motion state of a vehicle is presented.

Automotive radar interference mitigation using a sparse sampling approach

The application of radar sensors for driver assistance systems and autonomous driving leads to an increasing probability of radar interferences. Those interferences degrade the detection capabilities and can cause sensor blindness. This paper uses a realistic road scenario to address the problems of a common countermeasure that simply removes interference-affected parts of time domain radar signals and thereby introduces a gap. The paper solves the problem with the application of a sparse sampling signal recovery algorithm that is also used for compressed sensing problems. It is shown that the signal recovery can clearly overcome the shortcomings of just removing interfered signal parts. In the end of the paper, the applicability of the used algorithm is verified with measured radar data.