Research On Abnormal Driving Behaviors Supervision Based On Blockchain Incentive Mechanism

Online ride-hailing such as Didi has become one of the main way for people to travel.Ensuring safety and reducing traffic accidents are the core concerns of all online ride-hailing platforms.Abnormal driving behaviors such as fatigue driving and distracted driving are important factors that cause traffic accidents.At present,the solution to avoid fatigue driving on the ride-hailing platform is mainly to limit the driver’s driving time,and stop dispatching orders to the driver after a specified time.This one-size-fits-all solution doesn’t develop different regulatory measures based on the specific circumstances of each driver.For distracted driving,those platforms haven’t taken effective measures to supervise them.In response to the above problems,this thesis proposes an abnormal driving supervision scheme based on the blockchain incentive mechanism,and our work is shown as follows:(1)A multi-feature fusion method is proposed,which aggregates the driver’s blinking behavior and head posture during the trip to detect the driver’s driving state,and defines three driving states including normal driving,fatigue driving and distracted driving.(2)Driver’s personal credits rewards and punishments,incomes calculation rules based on blockchain and smart contracts to supervise the driver’s driving behaviors are proposed in this thesis.By triggering the smart contract deployed in the Ethereum blockchain to punish drivers with abnormal driving behaviors,personal credits will be punished,and drivers who maintain normal driving will be rewarded with personal credits.Finally,dynamically adjust the itinerary unit price according to the current credits of the personal account and calculate the order revenue.So as to realize the personalized supervision of the driver,and guide the driver to form good driving habits.(3)The Markov decision process is introduced,and the decision process of whether the driver continues to drive in abnormal driving conditions is abstracted as a Markov chain.The Markov decision process is used to simulate the driver’s driving habits and find reasonable driver rewards and punishments.Provide a quantifiable objective evaluation method for the driver’s rewards and punishments,thereby verifying the effectiveness of the incentive mechanism.This thesis integrates behavior detection,blockchain and smart contracts,and then applied those technologies to the field of driver behavior supervision to achieve effective supervision and incentives for driver behaviors.Our scheme improve safety and promote the harmonious development of online car-hailing services.