Autonomous driving predictive defensive driving system through interaction based on forward vehicle driving and situation judgement information
The present disclosure relates to a defensive driving vehicles interaction system (D2VIS), and to an autonomous driving predictive defensive driving system through an interaction based on forward vehicle driving and situation judgement information and a method thereof. An autonomous driving predictive defensive driving system through an interaction based on forward vehicle driving and situation judgement information according to the present disclosure includes an inter-vehicle distance recognition unit configured to recognize a distance between a surrounding vehicle and an ego vehicle, a situation recognition unit configured to recognize situation information including surrounding information of the surrounding vehicle, and a driving situation response determination unit configured to share data for determining a defensive driving action by using the situation information.
1 . An autonomous driving predictive defensive driving system through an interaction based on forward vehicle driving and situation judgement information, the autonomous driving predictive defensive driving system comprising a processor and memory configured to implement:
an inter-vehicle distance recognition unit configured to recognize a distance between a surrounding vehicle and an ego vehicle;
a situation recognition unit configured to recognize situation information including surrounding information of the surrounding vehicle; and
a driving situation response determination unit configured to share data for determining a defensive driving action by using the situation information;
wherein the inter-vehicle distance recognition unit comprises:
an image acquisition module configured to acquire an image by photographing a rear of a forward vehicle and a front of a rear vehicle using a mono camera;
a mono depth conversion module configured to convert the image into mono depth to secure three-dimensional information;
a detection module configured to detect the rear of the forward vehicle and the front of the rear vehicle from the image and to secure information for vehicle type classification;
a tracking module configured to track detection information about the rear of the forward vehicle and the front of the rear vehicle in an original image domain;
a tracking information integration module configured to integrate tracking results of the tracking module in a mono depth-converted domain; and
an estimation module configured to estimate a speed and a direction of the forward vehicle and the rear vehicle by analyzing changes in depth over time;
wherein the situation recognition unit comprises:
a cargo recognition module configured to recognize a class of cargo protruding outside a cargo box or loading box area of a cargo vehicle;
a target definition module configured to collect motion information and state information of forward and rear objects from the inter-vehicle distance recognition unit, determine target information, analyze a location and an action of the target objects based on distance, speed, and direction information, and define a safety level of the targets;
an action recognition module configured to use forward and rear vehicle information output from the target definition module and to recognize an action associated with a dangerous situation after detecting the dangerous situation using motion events of the ego vehicle, and to share with the forward and rear vehicles a hazard level associated with a driving condition of the ego vehicle, collision time, overall length and width of the ego vehicle, type and braking distance of the ego vehicle, and motion state information of the ego vehicle in consideration of a driving state of the ego vehicle; and
a determination module configured to analyze information received from the action recognition module to analyze movements of the forward vehicle and surrounding vehicles;
wherein the driving situation response determination unit is configured to:
provide, to the forward vehicle and surrounding vehicles, sudden braking information of the ego vehicle and dangerous state information of a forward situation,
provide distance information with the forward and rear vehicles before tail lights of the ego vehicle are turned on due to brake operation of the ego vehicle, and
generate a driving strategy of the ego vehicle using state information of the forward vehicle, state information of the rear vehicle, distance information, and state information of the ego vehicle, and provide the driving strategy to the rear vehicle.
2 . The autonomous driving predictive defensive driving system of claim 1 , wherein the inter-vehicle distance recognition unit acquires location information by using a track ID for the surrounding vehicle, and tracks the surrounding vehicle in a domain.