Active safety system protecting from lateral impact in selected scenarios

Transcription

1 MIROSŁAW Marcin 1 DZIEWOŃSKI Tomasz 2 Active safety system protecting from lateral impact in selected scenarios INTRODUCTION Last few years show tremendous increase of engineering development in active safety area targeting in both avoidance or minimizing the effects of the crash. Apart from already nowadays standard ABS or EPS, more and more forward-looking sensor-based Advanced Driver Assistance Systems (ADAS) systems including adaptive cruise control and collision warning / avoidance / mitigation control becomes standard equipment of modern vehicles. Recent analyses [1] of some of new systems perspectives for upcoming decade in terms their influence on changing accidents patterns and introducing significant improvements in safety measures, confirms their effectiveness and applicability mainly for frontal/rear impact type collision scenarios. For side impacts, specially occurring inside cities on urban crossing effectiveness of current active systems seems to be limited. Direct objective of the presented research was to develop reliable lateral impact mitigation control system based on 4 radar sensors tested in PreScan virtual environment for typical crossing lateralimpact type scenarios. Conducted analyses proves the effectiveness of the developed base-line control system for most of the typical cases, but at the same time outline its limitation for high traffic condition. Strong sensitivity/dependency of the decision control algorithm to the high number of input signals makes the system not efficient for swift reactions leaving the subject challenge for further analyses. 1 SYSTEM DEVELOPMENT The base-line lateral-impact protection system has been modelled in MATLAB environment and simulated with a set of four generic RADAR sensors in a PreScan software. 5 typical types of lateralimpact scenarios were analysed under light or heavy traffic condition to test systems reliability and robustness. 1.1 Lateral impact definition and motivation. Lateral impact was defined when collision happens at side part of the vehicle with primary direction of force (PDOF) between 45 and 135 or 225 and 315 degrees [2]. It should be noted that this definition covers both pure 3 or 9 o clock collisions that ends up with thoracic injuries and oblique ones for which head injuries are predominant [3]. Current injury statistics confirm that lateral impacts are the most dangerous types of accidents in terms of fatalities and injury risks, e.g year in United States of America was 7250 fatalities in result of lateral collision, while 3914 deaths in head-on collision. Number of fatalities in T-bone collision raised from 679 to 850, because of more SUV cars participating, that are 33% more vulnerable to rollover effect. In Europe 25% fatalities in effect of hard injuries sustained in lateral impact. 1.2 System concept Developed system consist of 4 radars and the control module (Figure 1) taking into account spotted object s distance and velocity. It calculates Time to Collision (TTC) and asses its probability. Two specific values of TTC are vital for breaks activation decision unit: TTC=2s defining possible impact situation and TTC=0.6s used for inevitable impact situation. TTC can be increased 1 Politechnika Warszawska Wydział Mechaniczny Energetyki i Lotnictwa mmiroslaw@meil.pw.edu.pl 2 Politechnika Warszawska Wydział Mechaniczny Energetyki i Lotnictwa tomekn@meil.pw.edu.pl 7540

2 by braking and in the effect collision can be avoided. The system is set to starts braking when TTC is below 2s (warning actuation level) instead of 0.6s, to show possibility of avoidance with driver reaction. Distance/ velocity of spotted object SENSOR UNIT - based on 4 radars CONTROL MODULE DECISION UNIT - calculated TTC -calculates collision probability - breaks activation Fig. 1 Control Module scheme Simulation of the system is done in PreScan software environment used to simulate behaviour of vehicles on the road [4] (Figure 2). It allows for intuitive modelling of realistic scenarios, defining standard road users actors (cars, pedestrians with build-in simple dynamics etc.) taking into account and easy integrating of any own MATLAB additional modules and in combination with SIMULINK, PreScan creates possibility to simulate and test (vehicle reactions) in repeatable condition[5] [6]. Fig. 2: LKAS simulation done in PreScan Control system concept has been tested for Audi A8 model, available with basic library of models in PreScan. Vehicle has been equipped in Path-Follower - controller simulation the driver that follows the trajectory with desired velocity. Simplified dynamics model of the A8 has been used (Figure 3). Radar sensors are simulated with Technology independent sensor (TIS) modules set to wide pyramid. Frequency of the sensor is 50 Hz for every of two beams, beam range 150m. Four sensors have been set in every corner of the car. 7541

3 Fig. 3: PreScan Audi A8 dynamics configuration card 1.3 Simulation scenarios Five relevant lateral impact scenarios have been modelled to test effectiveness of the designed system. Each case is illustrated with the frame from simulation run. Black car usually represent the car equipped with mitigation control system while red shape indicates its position in case the system is not active and collision take place. T-Bone collision: The first case shows the vehicles driving from different directions at the X- crossing (Figure 4). The left vehicle equipped with the design system slows down and passes another car. Fig. 4: T-Bone Collision. 7542

4 Collision while turning into traffic: In the second scenario, the black vehicle with the system on board is trying to join into traffic from parking space on the shoulder (Figure 5). The system prevents the vehicle from a collision with yellow car driven way. Fig. 5: Collision while turning into traffic Collision during left turn: The third experiment shows the bending host car at the X-crossing to the left, while the striking vehicle going straight through the intersection on the same street (Figure 6). The vehicle slowed so that collisions do not occur. Fig. 6: Collision during left turn Collision while entering roundabout: The fourth example shows a ride on the roundabout (Figure 7). The system slowed down the vehicle so that it does not enter the roundabout when the other vehicle is on it, what would result in a collision. 7543

5 Fig. 7: Collision while entering roundabout Collision during Lane change: In the fifth scenario, in case of a sudden lane change by the host car (Figure 8) the system s reaction is not sufficient and there is a collision. Fig. 8: Collision during sudden lane change 2 SYSTEM VERIFICATION The mitigation system concept has been tested for all 5 scenarios taking into account small variations of initial conditions (changes of nominal velocities or changing the traffic density by multiplying spotted by the sensors surrounding objects). As a good verification of system s effectiveness apart from visualisation of collision/non collision cases the velocity of the controlled car has been registered. 7544

6 Case 1 T-Bone Results: In scenario with T-Bone collision, system was able to avoid collision by braking. It lowered the velocity of the host car and allow car from the right to drive through the X-crossing. Efficiency of the system was confirmed with both changed velocity conditions and increased traffic Case -2 Turning into traffic Description: In second scenario, host car wait on the park place till target car is near. Then host car starts moving. System should activate braking, before turn into traffic, to prevent from collision. The breaking system is set to TTC = 0.6s, as driver would ignore warning, about possibility of collision, with parked cars. After target vehicle will be in safe area, host car should starts moving again. Results: In turning to traffic scenario, host car was prevented from collision. Problem with false alarms appears while driving parallel to the parked cars. Host car is very near of them, and in moment when it is in the nearest position of rear bumper of target car, the TTC and range are low enough to turn on braking. 7545

7 Case 3 Turn left Description: The host car turns left on the crossing, while target car moves straight (Figure 6). System should decrease velocity of the host car and avoid collision Results: Velocity of the host car has been decreased in all scenarios. With nominal and changed conditions, the system allowed to avoid the collision. In scenario with increased traffic, Collision with car ahead has been avoided, but reaction of the car was not enough to avoid collision from the car on the left. Case 4 Roundabout Description: In scenario with collision during entering roundabout, the host car tries to not give a way to the car on the crossing. System should decrease velocity and allow target car to safety leave the roundabout, and avoid the collision Results: In all scenarios with roundabout the system allows to avoid collision. In scenario with increased traffic, system decreased speed only a little and allowed the Host car to enter roundabout, when it was safety. All cars can move on the crossing without unneeded interfere Case 5 Lane change Results Collision during rapid lane change was impossible to avoid by braking. Velocity has been decreased, but it was not enough. Test shows that car can travels parallel to the other moving car without activation of the braking system. 7546

8 CONCLUSIONS The simulation results shows the effectiveness of developed system, that protects from lateral impact. In nominal conditions and variations (with limited traffic) system is working properly. It helped to avoid collisions in four of the five cases. But during the higher traffic, system reacts with false alarms. The paper summarizes concept and results of evaluation of system based on the currently available sensors. PROSPECTS Although developed radar based lateral collision mitigation system concept proved to be effective for low and normal traffic condition it still generate too many false alarms in high traffic condition. Further work would focus on elimination or logical smart management of false alarms signals in high traffic to improve reliability of the system concept. It is also planned to add and evaluate steering wheel control to improve collision avoidance capabilities of current brake based system in e.g. lane-change manoeuvre. Another challenging perspective would be to implement and validate virtually built and tested system concepts in real-life environment. ACKNOWLEDGEMENT Authors thank the TNO Automotive Safety Solutions B.V and its Polish representative - SEARCH SC company- for the PreScan software, advice and support. This work has been supported by the European Union in the framework of European Social Fund through the Didactic Development Program of the Faculty of Power and Aeronautical Engineering of the Warsaw University of Technology. Abstract Last few years show tremendous increase of engineering development in active safety area targeting in both avoidance or minimizing the effects of the crash. Apart from already nowadays standard ABS or EPS systems, more and more often forward-looking sensor-based systems such as Advanced Driver Assistance Systems (ADAS) including adaptive cruise control and collision warning/avoidance/mitigation control systems becomes more often standard equipment of modern vehicles. Paper summarizes development and virtual testing approach of own universal lateral collision mitigation control system. As a virtual simulation platform the PreScan software package has been used. The control system has been tested in five relevant (typical) lateral impact scenarios taking into account various speed, manoeuvres and/or traffic conditions. System aktywnego bezpieczeństwa chroniący przed zderzeniem bocznym w wybranych scenariuszach Streszczenie Ostatnie lata pokazują ogromny wzrost w dziedzinie inżynierii aktywnych systemów bezpeiczeństwa zarówno wspomagających kierowcę, unikających kolizji lub minimalizujące jej skutki. Oprócz obecnie standardowych systemów takich jak ABS I ESP, coraz częściej pojawiają się pojazdy wyposażone w systemy oparte na czujnikach Advanced Driver Assistance Systems(ADAS), zawierające systemy kontroli prędkości (ACC) oraz systemy ostrzegania przed kolizją /unikania/łągodzenia jej skutków. Artykuł podsumowuje rozwój i wirtualne testy systemu chroniącego przed zderzeniem bocznym. Do przeprowadzenia wirtualnych testów został uzyty pakiet PreScan. Controler systemu został sprawdzony w pięciu przypadkach (typowych) kolizji bocznych biorących pod uwagę różne prędkości, manewry oraz natężenie ruchu. REFERENCES 1. Future Accident Scenarios involving Small Electric Vehicles.- Mats Y Svensson, Gian Antonio D Addetta, Anna Carlsson, Christian Ewald, Peter Luttenberger, Christian Mayer, Johan Strandroth, Ernst Tomasch, Andreas Gutsche, Jac Wismans, IRCOBI Conferenec 2014 paper - IRC

9 2. Factors influencing pediatric injury in side impact collisions. K.B. Argogast, E.K. Moll, S.D. Morris, J Trauma, 51 (2001), pp Occupant injury severity from lateral collisions: a literature review. Laberge-Nadeau C, Bellavance F, Messier S, Vzina L, Pichette F. J Safety Res. 2009;40(6): A Scenario Based Simulation Environment for Developing Intelligent Vehicle Systems M. Tideman, F. Leneman, S. de Hair - Buijssen, K. Kawaguchi, Proceedings of JSAE Annual Congress (spring). Yokohama, Japan Development Tools for Active Safety Systems: PreScan and VeHIL F. Hendriks, M. Tideman, R. Pelders, Vehicular Electronics and Safety (ICVES), 2010 IEEE International Conference 6. Future Testing of Active Safety Systems M. Tideman, V. Ruiz Garate, R. Bours, K. Kietlinski, SAE Internat Journ of Passenger Cars-Electronic Electrical Systems 3.2 (2010):