In accordance with one aspect of the present disclosure, a vehicle controls a vehicle includes: acquiring a direction and a distance value of the obstacle as position information of the obstacle based on radar data received through at least one of a plurality of reception channels corresponding to the angular resolution in the lateral direction of an obstacle detector, identifies a collision point that may collide with the obstacle based on the acquired position information of the obstacle, controls at least one of steering and braking based on the position information of the identified collision point; and when controlling the steering, acquires a collision avoidance margin distance value corresponding to the position information of the identified collision point, predicts the collision position based on the position information of the obstacle and the information detected by the velocity detector, acquires a distance value between the predicted collision position and the current position, acquires a movement distance value in the lateral direction based on the acquired distance value and a preset turning radius of the vehicle, acquires a steering angle based on the acquired movement distance value in the lateral direction and the acquired collision avoidance margin distance value and controls steering based on the acquired steering angle.

Achieving safety and natural automatic driving needs a control platform using intelligence (such as learning function and artificial intelligence), but it is difficult to ensure operations suited for the behavior of a vehicle by the output of intelligence. An automatic driving device according to the present invention includes a control program for inputting outside information and vehicle information, and outputting a target control value for a vehicle. The control program has a first program for generating a first target control amount on the basis of a dynamically changing algorithm (which outputs operations based on learning function or artificial intelligence), and a second program for generating a second target control amount on the basis of a prescribed algorithm (which outputs operations according to traffic rules or driving morals).

Systems and methods are provided for navigating an autonomous vehicle using reinforcement learning techniques. In one implementation, a navigation system for a host vehicle may include at least one processing device programmed to: receive, from a camera, a plurality of images representative of an environment of the host vehicle; analyze the plurality of images to identify a navigational state associated with the host vehicle; provide the navigational state to a trained navigational system; receive, from the trained navigational system, a desired navigational action for execution by the host vehicle in response to the identified navigational state; analyze the desired navigational action relative to one or more predefined navigational constraints; determine an actual navigational action for the host vehicle, wherein the actual navigational action includes at least one modification of the desired navigational action determined based on the one or more predefined navigational constraints; and cause at least one adjustment of a navigational actuator of the host vehicle in response to the determined actual navigational action for the host vehicle.

A method is disclosed to operate a motor vehicle using a driver assistance system. The driver assistance system includes, for at least one driver assistance system function, at least one criterion relating to a vehicle environment. The at least one criterion is defined for an adaptation of the driver assistance system function, a number of features of the environment of the vehicle are detected, a probability of the occurrence of at least one criterion is estimated on the basis of a combination of the detected features, and the driver assistance system function is adapted on the basis of the estimated probability.

A deviation avoidance apparatus includes: a boundary detection section that detects boundaries defining both edges in a width direction of a traveling path on which an own vehicle travels; a deviation prediction section that predicts that the own vehicle will deviate from the traveling path based on a travelling condition of the own vehicle that travels on the traveling path defined by the boundaries detected by the boundary detection section; an object detection section that detects an object that exists on one of the boundaries, the one of the boundaries being on a side where the own vehicle deviates from the traveling path, or outside of the one of the boundaries; a deviation avoidance section that commands, when the deviation prediction section predicts that the own vehicle will deviate from the traveling path, a travel control unit to have the own vehicle avoid deviating from the traveling path, the travel control unit controlling the travelling condition; and a command value adjustment section that adjusts, when the object detection section detects the object, a command value to be output from the deviation avoidance section to the travel control unit such that a maximum movement position in a case where the own vehicle moves to the side where the own vehicle deviates from the traveling path is on an inside of the traveling path with respect to the one of the boundaries on the side where the own vehicle deviates from the traveling path.

In one embodiment, a system receives a captured image perceiving an environment of an ADV from an image capturing device of the ADV, where the captured image identifies an obstacle in motion near the ADV. The system generates a feasible area surrounding the moving obstacle based on a projection of the moving obstacle. If the ADV is within the feasible area, the system determines an upper bound velocity limit for the ADV. The system generates a trajectory having a trajectory velocity less than the upper bound velocity limit to control the ADV autonomously according to the trajectory such that if the ADV is within the feasible area the ADV is to decelerate.

Systems and methods are provided for distributing a crowdsourced sparse map for autonomous vehicle navigation. In one implementation, a method of generating a road navigation model for use in autonomous vehicle navigation may include receiving navigation information associated with a common road segment from a plurality of vehicles; storing the navigation information associated with the common road segment; generating at least a portion of an autonomous vehicle road navigation model for the common road segment based on the navigation information; and distributing the autonomous vehicle road navigation model to one or more autonomous vehicles for use in autonomously navigating along the common road segment. The autonomous vehicle road navigation model may include at least one line representation of a road surface feature extending along the common road segment, and each line representation may representing a path along the common road segment substantially corresponding with the road surface feature.

A vehicle situation determination device includes an input unit and a controller. The input unit receives information about a recognition result of recognizing one or a plurality of moving objects existing in a sidewalk region ahead of a vehicle in an advancing direction. The controller determines, based on the recognition result, that the vehicle is allowed to enter a passing scheduled region in a time period of a sparse state when a transition is made from the sparse state into a dense state. The sparse state is a state where density of the one or plurality of moving objects existing in the passing scheduled region is lower than or equal to a predetermined value. The dense state is a state where the density is higher than the predetermined value.

Systems and methods for autonomous vehicle control are disclosed herein. According to some implementations, a method includes a scenario-specific operation control evaluation module (SSOCEM) based on a route of the vehicle. The SSOCEM includes a preferred model and one or more fallback models that respectively determine candidate vehicle control actions. The method includes instantiating a SSOCEM instance based on the SSOCEM. The SSOCEM determines a candidate vehicle control action by determining an approximate amount of time needed to determine a solution to the preferred model and determining an approximate amount of time until the upcoming scenario is reached. When the approximate amount of time needed to determine the solution is less than the approximate amount of time to reach the upcoming scenario, the candidate vehicle control action is determined based on the preferred model; otherwise, the candidate vehicle control action is determined based on a fallback model.

Systems and methods of deep neural network based parking assistance is provided. A system can receive data sensed by one or more sensors mounted on a vehicle located at a parking zone. The system generates, from a first neural network, a digital map based on the data sensed by the one or more sensors. The system generates, from a second neural network, a first path based on the three-dimensional dynamic map. The system receives vehicle dynamics information from a second one or more sensors located on the vehicle. The system generates, with a third neural network, a second path to park the vehicle based on the first path, vehicle dynamics information and at least one historical path stored in vehicle memory. The system provides commands to control the vehicle to follow the second path to park the vehicle in the parking zone.