COMPOSE develops and evaluates collision mitigation and vulnerable road user protection systems for trucks and cars. COMPOSE focuses on pre-crash measures to be taken during the last second before a collision. Various application measures are used: mitigation through autonomous braking / brake assist, and also utilization of protective structural actuators. Various innovative sensor data fusion methods will be pursued to achieve the high degree of sensor reliability, which is mandatory for such applications. (BMW rather than TU Munich is involved in this project.)
Within the PReVENT Integrated Project, the three-year COMPOSE subproject develops, integrates and validates a pre-crash safety application for collision mitigation by autonomous or semi-autonomous braking. This application reduces significantly the kinetic energy at the impact by mutual adaptation.
COMPOSE focuses on the following challenges:
Because of the typical human reaction time of 1 s or more, a driver warning strategy is not applicable during the last period of time preceding a collision. Rather, only autonomous systems can cope with collision mitigation and road user protection during the pre-crash period. The latter can be divided into three sub-intervals:
(i) The period between a few seconds (1-3 s, typically) and about 100 ms preceding the initial impact,
(ii) the pre-fire phase which lasts from about 100 ms to 10 ms before the crash and
(iii) the pre-set phase which covers roughly the last 10 ms before an encounter.
COMPOSE focuses on period (i). In order to achieve these goals, some machine perception system has to be developed which offers obstacle detection and localisation capabilities of an outstanding degree of reliability and robustness, even under adverse weather and illumination conditions. This is a prerequisite for the novel, preventive safety systems, as it ensures the avoidance of false alarms and the related collateral damages. From the technical and scientific point of view the perception reliability will be accomplished by various novel sensor data fusion strategies that take advantage of the redundancy and complementarity of a suitable set of presently available sensor devices.
![]() ![]() ![]() |
L. Walchshäusl, R. Lindl, K. Vogel, T. Tatschke
Detection of Road Users in Fused Sensor Data Streams for Collision Mitigation 10th International Forum on Advanced Microsystems and Automotive Applications, Berlin, Germany, Apr. 25-27 2006 (bib) |