Michael Heuer

Detection and tracking approach using an automotive radar to increase active pedestrian safety

Vulnerable road users, e.g. pedestrians, have a high impact on fatal accident numbers. To reduce these statistics, car manufactures are intensively developing suitable safety systems. Hereby, fast and reliable environment recognition is a major challenge. In this paper we describe a tracking approach that is only based on a 24 GHz radar sensor. While common radar signal processing loses much information, we make use of a track-before-detect filter to incorporate raw measurements. It is explained how the Range-Doppler spectrum of pedestrian can help to initiate and stabilize tracking even in occultation scenarios compared to sensors in series.

Requirements on automotive radar systems for enhanced pedestrian protection

This paper work is connected to the joint EU-project called ARTRAC. It briefly describes one of the first project tasks in which the requirements are derived that are necessary for enhanced pedestrian protection for a single radar system. Therefore, we examine different points of view: accident statistics and the associated environmental circumstances, user needs for active safety systems and radar regulations. The resulting technical requirements are compared to current automotive radar systems, while reviewing the feasibility. In addition we discuss how a new transmit/ receive antenna and multi-channel receiver should be developed and give examples for possible safety functions that could benefit from this innovative radar sensor.

Multi-modal fusion with particle filter for speaker localization and tracking

This paper describes a methodology for fusing multi-modal data meaningful together, in order to detect and track a speaker with a conventional sensor setup. We use Gaussian mixtures to combine the sensor information within a particle filter, such that a single speaker can be identified in the presence of multiple visual observations. The major advantages are design considerations that let the system perform in real time, while using an easily extensible framework. Besides, we highly reduce noise which gives us a more dependable prediction. Results illustrate the localization estimations in a two- and a three-person scenario.

Machine vision based recognition of emotions using the circumplex model of affect

In the vision community recognition and analysis of facial expressions is a long investigated area of research with potential applications such as human computer interaction (HCI). Most state of the art techniques focus on recognition of defined facial expression categories, i.e. basic emotions. However, in HCI classical basic emotions are only occurring sparsely, thus they are inadequate to guide the dialog with the user. Here we suggest a mapping to the so called circumplex model of affect, known from psychology. This enables us to determine the current affective state of the user, which can then be used to control the course of the interaction. In particular, the output of the proposed machine vision based recognition method gives insight to the observed persons arousal state and the valence along two axes of the 2-D plane of the underlying circumplex model. In this paper we present the basic principles of our new method and present first results.

Speaker tracking using multi-modal fusion framework

This paper introduces a framework by which multi-modal sensory data can be efficiently and meaningfully combined in the application of speaker tracking. This framework fuses together four different observation types taken from multi-modal sensors. The advantages of this fusion are that weak sensory data from either modality can be reinforced, and the presence of noise can be reduced. We propose a method of combining these modalities by employing a particle filter. This method offers satisfied real-time performance.

Pedestrian tracking with occlusion using a 24 GHz automotive radar

Vulnerable road users, e.g. pedestrians, have a high impact on fatal accident numbers. To reduce these statistics, car manufactures are intensively developing suitable safety systems. Hereby, fast and reliable environment recognition is a major challenge. In this paper we describe a tracking approach that is only based on a 24 GHz radar sensor. While common radar signal processing loses much information, we make use of a track-before-detect filter to incorporate raw measurements. It is explained how the Range-Doppler spectrum can help to indicated pedestrians and stabilize tracking even in occluded scenarios compared to sensors in series.

An Active Vulnerable Road User Protection Based on One 24 GHz Automotive Radar

Among other initiatives to improve safety of Vulnerable Road Users (VRUs), the European Commission is funding a research project called ARTRAC: “Advanced Radar Tracking and Classification for Enhanced Road Safety” aimed to develop an integrated safety concept for pedestrians. ARTRAC investigates the possibility to use a stand-alone radar sensor especially for the detection of VRUs in order to generate the appropriate supporting actions to the driver with the goal to avoid or mitigate a possible collision. The supporting actions comprise driver warning, automatic braking and steering recommendation. This paper provides an overview of the definition of the system solutions for the warning and vehicle automatic actuations in case of a potentially dangerous situation with respect to a VRU occurs; and of the development of specific software for the actuation over the vehicle braking and steering with the goal to perform an automatic response devoted to avoid an accident involving a VRU or, in the worst case, to mitigate the impact. The functional performance of the systems is analyzed with a specialized pedestrian test facility.