Road accidents, with their human and economic costs, are a serious problem in most industrialised countries. For example, in year 2000 in the European Community 40000 people died and 1.7 million were injured, because of road accidents.
The project dealt with the development of drive/ride assistance systems for preventive active safety. These advanced devices monitor the behaviour of the vehicle, estimate the risk of the actual driving conditions by comparing the planned manoeuvre with a reference 'safe' manoeuvres and provide warnings to the driver/rider if the risk level crosses a threshold. They should prevent accidents caused by distractions/careless driving, by excessive speed or unsafe headway.
The system architecture is made of three large functional blocks, named 'Scenario and Navigation System', 'Risk Assessment and Safe Manoeuvre Calculation' and 'Driver Feedback'.
In 2004 there were many ongoing and starting research projects on this topic, but most research projects (especially those driven by automotive industries) had short to medium time frames and aimed at early deployment of marketable products. In addition they focused mainly on cars, which represent the great majority of vehicles causing accidents. Conversely the present project aimed at developing basic research on a novel 'safe'-optimal manoeuvre planning core, as well as to address the second most important, yet neglected, category of vehicles, which are motorcycles. The two aspects (safe-optimal manoeuvre core and motorcycles) are interconnected, because motorcycles, due to their more complex dynamics, pose planning problems that are more challenging than cars.
The main aim of the project was to develop a demonstrator for ride assistance which rates risk during the course and provides warning feedback to motorcyclist.
The most important objectives are the following:
- mathematical definition of risk (risk function) by comparing an actual manoeuvre with a 'safe' manoeuvre
- development of an improved solver for the safe-optimal manoeuvres suitable for real-time applications
- development of systems supporting the riding task that warn the rider of the risk by means of vibrations
- experimental testing of the systems supporting the riding task by means of a riding simulator and full-scale motorcycle dynamics.
The results achieved in the framework of this research programme will be the base of knowledge which is necessary in order to implement systems supporting the riding task in actual two-wheeled vehicles, in the short term and to develop next generation drive/ride assistance systems for all vehicles in the medium-long term.
The methodology consists of:
- road tests on motorcycles equipped with specific sensors in order to capture the most important features of safe manoeuvres;
- numerical modelling with multi-body codes for simulating motorcycle dynamics in riding simulators;
- optimal control methods for the definition of the safe manoeuvre and the calculation of the risk function;
- virtual reality techniques, vision systems, advanced actuation systems to improve the riding sensations given by the simulator;
- advanced test benches for the analysis of the dynamic characteristics of two-wheeled vehicles.
1. Characterisation of actual safe manoeuvres.
This activity has been carried out experimentally by means of motorcycles equipped with sensors and data loggers. The main results were the exact identification of the trajectory of the vehicle, vehicle attitude and actions carried by the rider during safe operations. The availability of telemetric data of actual manoeuvres resulted in being very useful for the set-up and tuning of the riding simulator.
2. Definition of risk function and software implementation.
In this activity the risk factors have been determined and a specific mathematical formulation has been developed for making risk a measurable quantity (a number) and giving a rigorous definition of manoeuvre safety.
Risk is defined as the integral, from actual time (t) to the assumed time horizon (t+T), of the 'instantaneous risk', weighted according to function which gives more importance to risky events in the nearest future and decreasing less importance to the same events further in time (because more time is left to take corrective actions).The 'instantaneous risk' itself accounts at least for the cost of using tyre forces, and for the cost of clearance with obstacles. Fixed obstacles are treated different from obstacles that may move. The risk associated with passing close to fixed obstacles depends only on the clearance left. Conversely the risk of obstacles that may move considers also the fact that clearance may be reduced because of possible acceleration of the obstacle itself.
Then a specific code for assessing the level of risk during the execution of a manoeuvre has been developed. The expression of risk level depends on several parameters in order to adjust the relative importance of different risk factors, and to adapt to different environmental conditions, to different vehicles and possibly even to personal preferences.
3. Development of real time software
The time needed for solving the optimal control problems depends on the complexity of the mathematical model of the vehicle and environment. At the beginning of the project the planning was already suitable for the simplest applications, e.g. a simple car model limited to longitudinal control. Since the dynamics of a two wheeled vehicle is much more complex, a real time code (named Xoptima) for the solution of optimal control problem has been developed. The model includes 10 equa
Recommended topics for further research are:
- definition of procedures for tuning of motorcycle riding simulators;
- definition of risk functions suited to the application in drive/ride assistance systems;
- definition of the main features on warning systems based on the transmission of vibration to the rider;
- definition of testing protocols to evaluate on the road the manoeuvres of two-wheeled vehicles.