Final Report Summary - ABRAM (Autonomous BRAking for Motorcycles)
Mobility is vital for the quality of life and efficient transport is an important goal for society. An increased use of motorcycles and mopeds - also known as powered two wheelers (PTWs) - instead of passenger cars for the needs of personal mobility would result in more efficient utilisation of road and parking space, less pollution, and time savings. However, the risks of being seriously injured or killed are higher for PTW riders than for other motorized road users. In particular, the European Road Safety Observatory (ERSO) indicated that the risk of incurring into a fatal crash while riding a PTW per km travelled is 12 times higher than for driving a car. In the last five decades passenger cars have benefitted of many new safety technologies, some of which at a later stage were translated to motorcycles – occasionally with proven success, for example antilock brake system. At present, one of the most advanced safety systems available is Autonomous Emergency Braking (AEB). AEB is designed to recognise an impending collision and then apply brakes when the driver does not. AEB effectiveness was recently demonstrated for cars, and AEB is promoted as safety feature by NCAP and a range of government bodies. However, AEB is currently not available for motorcycles and its possible applicability is unknown. In this context, the ABRAM project (Autonomous Braking for Motorcycles, June 2013-May 2016) proposed a broad exploration of AEB applied to PTWs. ABRAM addressed crucial aspects of the evaluation and implementation of autonomous braking for PTWs in order to create scientific bases for a possible development of this safety system.
The principal investigator of ABRAM was Dr Giovanni Savino. The first phase of the project (period May 2013- April 2015) took place at the Monash University Accident Research Centre (MUARC), Australia. MUARC is recognised internationally as a leading research centre in the field of road safety and has an excellent track record in projects dealing with the evaluation of safety technologies for vehicles. The second phase of the project (period May 2015 - May 2016) was conducted at the Department of Industrial Engineering of the University of Florence, Italy, with the research group MOVING led by Prof Marco Pierini.
The project was divided in seven work packages (WPs).
In WP1, a literature review of the injury risk factors for motorcycle riders was conducted.
WP2 focused on the development of an idealised AEB for motorcycles. It considered the applicability of AEB to real-world crashes; triggering algorithms were developed and tuned for typical motorcycle crash scenarios. Potential benefits of AEB for motorcycles were then evaluated via computer simulations of real world crash cases from three different countries: Italy, Sweden and New South Wales (Australia).
WP3 analysed the feasibility of a mild, unexpected automatic deceleration of the motorcycle, such as the one produced by AEB, from the viewpoint of the rider. An experimental study using an instrumented test motorcycle equipped with a special automatic braking system was conducted at the low speed test track of Bosch Australia. The trials involved the activation of the automatic braking system while participants were riding the test vehicle at around 40 km/h along a straight.
WP4 focused on the possible interactions between AEB intervention and rider control actions during the pre-crash phase. A low-cost motorcycle riding simulator was developed to measure realistic steering inputs of riders during emergency. The simulator was then used in study the behaviour of riders when facing AEB activation scenarios in a virtual environment.
In WP5, a sensitivity analysis of the effects of a motorcycle AEB system was proposed. The aim was to assess the validity of estimates of the effects that AEB may have produced in real world motorcycle crashes. These effects are typically evaluated using computer simulations based on case reports of in-depth crash investigations, in which some degree of uncertainties is inevitable. The activity also compared the potential benefits of a realistic AEB systems (based on current technology) and those of an ideal AEB system (with unlimited field of field of view and zero delays on obstacle detection/identification).
WP6 was an on-road test of a current sensing technology for obstacle detection applied to a test motorcycle. This technology, a laser scanner sensor, was used to detect vehicles in front of the motorcycle during normal riding in the traffic of Florence. The test showed current capabilities and limitations of the sensing technology, in a comparison with the performances required by the triggering algorithms defined in WP2.
WP7 dealt with estimating the potential benefits of AEB for motorcycles in comparison with other technologies, in terms of societal costs that could be saved by avoiding or mitigating crashes. The activity used historical motorcycle crash data in Victoria, Australia.
Results of this research showed that AEB has a potential for an application to real world motorcycle crashes deserving further consideration. The first outcome was the definition of new triggering algorithms for AEB designed to deploy an automatic deceleration as soon as a collision with an opponent vehicle becomes physically inevitable (that is to say, no matter what manoeuvre is executed by the host motorcycle's rider and by the opponent vehicle's driver). The proposed triggering algorithms apply to any pre-crash scenario involving two vehicles and were arranged in the form of look up tables to allow for easy implementations in embedded systems. The second outcome was an evaluation of the potential applicability of such system. Notwithstanding the restrictions needed for a safe AEB deployment on motorcycles (bike travelling along a straight, obstacle in the field of view, rider not attempting a sharp turn before the collision), estimated applicability of AEB ranged from 28% to 32% of the sample motorcycle crashes used in the study. Impact speed reductions operated by AEB were up to 7 km/h in the computer simulations. These results were proven to be generally robust against uncertainties of the crash reconstructions and against limitations of currently available technologies. Impact speed reductions that AEB may achieve are expected to produce an injury mitigation effect for the riders. However, an evaluation of quantitative benefits in terms of injury reduction has not been part of the present activity and represents one of the priorities of future studies. The ABRAM project also analysed a variant approach to AEB for motorcycle-to-car crashes. AEB deployment on a motorcycle is much more critical than AEB on a passenger car. With this in mind, triggering algorithms were modified to deploy full braking of the passenger car as soon as a collision between a motorcycle and that passenger car becomes imminent (still avoidable though), and at the same time a collision can still be avoided only via extreme, non ordinary manoeuvres (such as emergency braking or emergency swerving). This approach assumed that an exchange of information between vehicles is in place (motorcycle to car communication), Simulations of the crash cases analysed for the purposes of motorcycle AEB were repeated with this new triggering algorithms of a collaborative AEB for passenger cars. Results showed that 50% of the considered crashes could have been avoided by fully braking the passenger car, with no automatic action on the motorcycle. Automatic full braking of a car is an application already implemented on the market, however little information is available concerning the feasibility of automatic decelerations on a motorcycle. In this regard, ABRAM tests proved that standard riders can sustain a moderate automatic deceleration (15-20% of full braking on dry asphalt, such as the deceleration produced by cutting the engine ignition) with minor-to-moderate efforts, even when the activation is unexpected. Further investigation is still required to explore the reactions of the rider in the imminence of a collision, with and without the intervention of an AEB system. In this context, the contribution of the ABRAM project is the definition and initial validation of a low-cost, easy-to-implement motorcycle simulator rig for the analysis of riders' reactions when facing pre-crash scenarios.
A dissemination of the ABRAM research activities and results was done via the project blog page: www.abram-project.blogspot.com and via Dr Savino’s professional Twitter account (gioptw).
The principal investigator of ABRAM was Dr Giovanni Savino. The first phase of the project (period May 2013- April 2015) took place at the Monash University Accident Research Centre (MUARC), Australia. MUARC is recognised internationally as a leading research centre in the field of road safety and has an excellent track record in projects dealing with the evaluation of safety technologies for vehicles. The second phase of the project (period May 2015 - May 2016) was conducted at the Department of Industrial Engineering of the University of Florence, Italy, with the research group MOVING led by Prof Marco Pierini.
The project was divided in seven work packages (WPs).
In WP1, a literature review of the injury risk factors for motorcycle riders was conducted.
WP2 focused on the development of an idealised AEB for motorcycles. It considered the applicability of AEB to real-world crashes; triggering algorithms were developed and tuned for typical motorcycle crash scenarios. Potential benefits of AEB for motorcycles were then evaluated via computer simulations of real world crash cases from three different countries: Italy, Sweden and New South Wales (Australia).
WP3 analysed the feasibility of a mild, unexpected automatic deceleration of the motorcycle, such as the one produced by AEB, from the viewpoint of the rider. An experimental study using an instrumented test motorcycle equipped with a special automatic braking system was conducted at the low speed test track of Bosch Australia. The trials involved the activation of the automatic braking system while participants were riding the test vehicle at around 40 km/h along a straight.
WP4 focused on the possible interactions between AEB intervention and rider control actions during the pre-crash phase. A low-cost motorcycle riding simulator was developed to measure realistic steering inputs of riders during emergency. The simulator was then used in study the behaviour of riders when facing AEB activation scenarios in a virtual environment.
In WP5, a sensitivity analysis of the effects of a motorcycle AEB system was proposed. The aim was to assess the validity of estimates of the effects that AEB may have produced in real world motorcycle crashes. These effects are typically evaluated using computer simulations based on case reports of in-depth crash investigations, in which some degree of uncertainties is inevitable. The activity also compared the potential benefits of a realistic AEB systems (based on current technology) and those of an ideal AEB system (with unlimited field of field of view and zero delays on obstacle detection/identification).
WP6 was an on-road test of a current sensing technology for obstacle detection applied to a test motorcycle. This technology, a laser scanner sensor, was used to detect vehicles in front of the motorcycle during normal riding in the traffic of Florence. The test showed current capabilities and limitations of the sensing technology, in a comparison with the performances required by the triggering algorithms defined in WP2.
WP7 dealt with estimating the potential benefits of AEB for motorcycles in comparison with other technologies, in terms of societal costs that could be saved by avoiding or mitigating crashes. The activity used historical motorcycle crash data in Victoria, Australia.
Results of this research showed that AEB has a potential for an application to real world motorcycle crashes deserving further consideration. The first outcome was the definition of new triggering algorithms for AEB designed to deploy an automatic deceleration as soon as a collision with an opponent vehicle becomes physically inevitable (that is to say, no matter what manoeuvre is executed by the host motorcycle's rider and by the opponent vehicle's driver). The proposed triggering algorithms apply to any pre-crash scenario involving two vehicles and were arranged in the form of look up tables to allow for easy implementations in embedded systems. The second outcome was an evaluation of the potential applicability of such system. Notwithstanding the restrictions needed for a safe AEB deployment on motorcycles (bike travelling along a straight, obstacle in the field of view, rider not attempting a sharp turn before the collision), estimated applicability of AEB ranged from 28% to 32% of the sample motorcycle crashes used in the study. Impact speed reductions operated by AEB were up to 7 km/h in the computer simulations. These results were proven to be generally robust against uncertainties of the crash reconstructions and against limitations of currently available technologies. Impact speed reductions that AEB may achieve are expected to produce an injury mitigation effect for the riders. However, an evaluation of quantitative benefits in terms of injury reduction has not been part of the present activity and represents one of the priorities of future studies. The ABRAM project also analysed a variant approach to AEB for motorcycle-to-car crashes. AEB deployment on a motorcycle is much more critical than AEB on a passenger car. With this in mind, triggering algorithms were modified to deploy full braking of the passenger car as soon as a collision between a motorcycle and that passenger car becomes imminent (still avoidable though), and at the same time a collision can still be avoided only via extreme, non ordinary manoeuvres (such as emergency braking or emergency swerving). This approach assumed that an exchange of information between vehicles is in place (motorcycle to car communication), Simulations of the crash cases analysed for the purposes of motorcycle AEB were repeated with this new triggering algorithms of a collaborative AEB for passenger cars. Results showed that 50% of the considered crashes could have been avoided by fully braking the passenger car, with no automatic action on the motorcycle. Automatic full braking of a car is an application already implemented on the market, however little information is available concerning the feasibility of automatic decelerations on a motorcycle. In this regard, ABRAM tests proved that standard riders can sustain a moderate automatic deceleration (15-20% of full braking on dry asphalt, such as the deceleration produced by cutting the engine ignition) with minor-to-moderate efforts, even when the activation is unexpected. Further investigation is still required to explore the reactions of the rider in the imminence of a collision, with and without the intervention of an AEB system. In this context, the contribution of the ABRAM project is the definition and initial validation of a low-cost, easy-to-implement motorcycle simulator rig for the analysis of riders' reactions when facing pre-crash scenarios.
A dissemination of the ABRAM research activities and results was done via the project blog page: www.abram-project.blogspot.com and via Dr Savino’s professional Twitter account (gioptw).
