Marc M. Muntzinger

Probability estimation for an automotive Pre-Crash application with short filter settling times

In this paper, the merits of incorporating covariance propagation into a real-time Pre-Crash application are investigated. The suggested Pre-Crash algorithm activates restraint systems, such as a reversible seat belt tightening system, before an unavoidable accident happens. Sensor fusion of two short-range and one long-range radar with a target-based fusion is used to realize this vehicle safety application. A powerful, yet applicable method for using not only state but also covariance information for triggering actuators is proposed. A comprehensive parameter study on simulated as well as on real data shows statistically significant improvements in detection rate. Further, the importance of covariance errors in terms of accuracy for Pre-Crash applications is demonstrated. Even with few detection cycles and short filter settling times, a good compromise between detection rate and false alarms can be deduced.

Making Bertha See Even More: Radar Contribution

For decades, radar has been applied extensively in warfare, earth observation, rain detection, and industrial applications. All those areas are characterized by requirements such as high quality of service, reliability, robustness in harsh environment and short update time for environmental perception, and even imaging tasks. In the vehicle safety and driver assistance field, radars have found widespread application globally in nearly all vehicle brands. With the market introduction of the 2014 Mercedes-Benz S-Class vehicle equipped with six radar sensors covering the vehicles environment 360° in the near (up to 40 m) and far range (up to 200 m), autonomous driving has become a reality even in low-speed highway scenarios. A large azimuth field of view, multimodality and a high update rate have been the key innovations on the radar side. One major step toward autonomous driving was made in August 2013. A Mercedes-Benz research S-Class vehicle-referred to at Mercedes as Bertha-drove completely autonomously for about 100 km from Mannheim to Pforzheim, Germany. It followed the well-known historic Bertha Benz Memorial Route. This was done on the basis of one stereo vision system, comprising several long and short range radar sensors. These radars have been modified in Doppler resolution and dramatically improved in their perception capabilities. The new algorithms consider that urban scenarios are characterized by significantly shorter reaction and observation times, shorter mean free distances, a 360° interaction zone, and a large variety of object types to be considered. This paper describes the main challenges that Daimler radar researchers faced and their solutions to make Bertha see.

Impact of out-of-sequence measurements on the joint integrated probabilistic data association filter for vehicle safety systems

This paper addresses the problem of joint state and existence estimation in the presence of temporally asynchronous measurements. In multi-sensor fusion, the problem can occur that measurements from different sensors can arrive at the processing unit out of sequence, i.e., the original temporal order of the measurements is lost. For the first time, the influence of these out-of-sequence measurements on state estimation as well as on existence estimation is examined. The existence probabilities are estimated via the Joint Integrated Probabilistic Data Association Filter (JIPDA) [17]. Two different methods to deal with out-of-sequence measurements in JIPDA are described and compared. It is shown that the handling of out-of-sequence measurements has a considerable influence not only on state, but also on existence estimation.

Tracking in a cluttered environment with out-of-sequence measurements

Recent literature has introduced several methods for processing an out-of-sequence measurement (OOSM) in tracking for both the 1-step-lag and ¿-step-lag case. However, in a realistic tracking application, a data association algorithm, such Joint Probabilistic Data Association (JPDA) must be used in order to associate measurements with tracks in a cluttered environment. This paper investigates the OOSM problem in tracking for both the 1-step-lag and ¿-step-lag cases using JPDA. The OOSM algorithms presented in the literature are extended to support JPDA and the results are analyzed. The resulting algorithms are then applied to an automotive frontal pre-crash system, where OOSM and JPDA are vital factors in improving the performance of such a system.

Out-of-sequence measurement processing for an automotive pre-crash application

In this paper, the merits of incorporating out-of-sequence measurements (OOSM) into a Pre-Crash application are investigated. When an imminent front crash is detected by the Pre-Crash system, the algorithm activates a reversible seat belt tightening system. This paper points out that simple buffering is not applicable in most time critical applications such as Pre-Crash. It is crucial to have sufficiently accurate tracking information without any buffering delays, especially in urban traffic scenarios. Furthermore, the existing OOSM algorithm from Bar-Shalom [1] for the 1-step-lag case is extended to support Joint Probabilistic Data Association (JPDA). A comprehensive evaluation on simulated as well as on real sensor data is presented.

Present research activities and future requirements on automotive radar from a car manufacturer??s point of view

Automotive Radar has already found its way into nearly all car manufacturers portfolio, even for small car platforms. Over the decades, the performance requirements increased steadily from simple detector tasks in blind spot monitoring systems to multi range smart environment perception sensors. The utmost push in performance requirement is initiated with the trend towards highly automated driving. Future automotive radar systems have to provide imaging like capabilities and have to interact in radar networks, which allow for highly comprehensive perception tasks. The paper will provide an overview on state of the art automotive radar usage on the basis of the DAIMLER car platforms, will give an outline on future requirements for highly automated driving and will present recent approaches in radar based environmental perception.

Multi-sensor fusion with out-of-sequence measurements for vehicle environment perception

Automated driving applications require an environment perception that is reliable and fast. Multi-sensor fusion is a suitable means to combine the advantages of different measurement principles. However, this may lead to out-of-sequence measurements, i.e., asynchronous measurements where the original order of the measurements is lost. High-performance out-of-sequence algorithms are therefore needed that do not depend on the order of the measurements. In addition, existence probabilities can increase the reliability of the fusion system especially in safety critical applications. This paper presents a novel approach to handle out-of-sequence measurements not only in state estimation, but also in existence estimation. The method is shown to result in equal or less computational costs than state-of-the-art methods. The proposed algorithm is evaluated with real world data from crash tests.

State and existence estimation with out-of-sequence measurements for a collision avoidance system

As their functionality becomes more and more complex, future driver assistance systems rely on several different sensors in order to combine the advantages of different measurement principles. However, in multi-sensor fusion, measurements may arrive at the fusion unit out-of-sequence, the original order of the measurements may be lost. Whereas out-of-sequence measurement processing in state estimation has been studied extensively, their incorporation in existence estimation has not been solved in the past. This paper presents a new algorithm for state and existence estimation in time-critical applications, where out-of-sequence measurements are handled adequately. The derived algorithm is validated with real-world data from crash tests.

Multi-target tracking for merged measurements of automotive narrow-band radar sensors

In this paper, the conventional JPDAM (joint probabilistic data association for merged measurements) algorithm for modelling merged observations, is applied to the automotive environment. As every sensor has limited resolution, the JPDAM should be used instead of the JPDA algorithm, which does not model the effects of measurement merging. As the JPDAM algorithm is more complicated, these effect are normally neglected for most sensors with good resolution capability. In this paper a new automotive prototypical multi-beam monopulse narrow-band FMCW radar sensor is used. While this sensor has good detection accuracy, it has only limited resolution capability. For this sensor, measurement merging leads to wrong target estimations when not modelled correctly in the multi-target tracking process. While the JPDAM algorithm has already been developed for this case, it can not be applied directly, as it does not model the diminished measurement accuracy when more than one measurement is in the same resolution cell. When using a gating procedure with standard measurement accuracy, the merged measurements will not be associated to the corresponding targets. This motivates the usage of a modified gating method which will be presented in this paper. To reduce the computational demands of the algorithm, a clustering technique for the modified gating procedure is shown. The modified algorithm is tested with real sensor data. The JPDA algorithm and the JPDAM algorithm are both applied to a challenging measurement merging scenario and the results of these algorithms are compared.

Time-to-Collision Estimation in Automotive Multi-Sensor Fusion with Delayed Measurements

This paper presents a time-to-collision estimation in the context of multi-sensor fusion. Several asynchronous sensors are fused where the measurements arrive at the fusion unit out-of-sequence, i.e., some measurements are temporally more delayed than others. The adequate out-of-sequence handling is crucial for time-critical applications such as pre-crash systems. Several methods are discussed and compared with respect to accuracy and computational costs. In addition, a reduced out-of-sequence algorithm for practical application is derived. The performance of the pre-crash system is evaluated using real-world data from crash tests. To this end, a soft crash target is used with a position ground truth accurate to the centimeter and a contact sensor as temporal ground truth.