Ford Global Technolgoies, LLC

Mohsen Lakehal-ayat

Jinhyoung Oh

Yi Zhang

Thomas Edward Pilutti

Eric Hongtei Tseng

Praveen Narayanan

Frank Peter Engels

James Carthew

Uwe Hoffmann

Muhammad Adeel Awan

Kaushik Balakrishnan

Weifeng Xiong

Saeid Nooshabadi

Zijian Wang

A method for training an autonomous vehicle to reach a target location. The method includes detecting the state of an autonomous vehicle in a simulated environment, and using a neural network to navigate the vehicle from an initial location to a target destination. During the training phase, a second neural network may reward the first neural network for a desired action taken by the autonomous vehicle, and may penalize the first neural network for an undesired action taken by the autonomous vehicle. A corresponding system and computer program product are also disclosed and claimed herein.

Automatic navigation using deep reinforcement learning

A trailer detection system for a vehicle includes a radar system outputting object point location data to a rear of the vehicle and a controller. The controller receives the object point location data and determines a position of an edge of the trailer relative to the vehicle by processing the object point location data from the radar system at a reduced threshold relative to use of the data in detecting a vehicle and filters out object point location data not correlated with the edge of the trailer based on object point location data persistence during driving of the vehicle. The controller further correlates the determined position of the edge of the trailer relative to the vehicle to determine an angle of the trailer relative to the vehicle about a coupling point.

Hitch angle detection using automotive radar

Exemplary embodiments described in this disclosure are generally directed to a multi-drone automotive system that includes a first drone configured to carry one or more detachable drones. The first drone, which may be referred to as a carrier drone, may be mounted upon an automobile and operated in a tethered mode of flight. The detachable drones may be launched from the carrier drone to carry out untethered flight. The carrier drone and/or the detachable drones may be used for various applications. In one example application, the carrier drone may use a first camera that is mounted upon the carrier drone, to capture a first set of images during the tethered mode of flight. A detachable drone may be launched from the carrier drone in an untethered mode of flight in order to capture a second set of images by using a second camera mounted upon the detachable drone.

Multi-drone automotive systems and methods of use

A system for control of a steering system of a vehicle, mechanically coupled to the wheels and including an actuator for applying a force or a torque to the steering system. A force or torque can be superimposed on a force or torque originating from the wheels. The system includes a detection unit disposed on the vehicle and configured for anticipatorily detecting at least one surface condition of a surface section located ahead of the vehicle in the direction of vehicle travel and subsequently driven on by the vehicle. The system including a data processing unit disposed on the vehicle and connected to and communicating with the detection unit. The data processing unit configured for generating control signals for controlling an actuator of the steering system based on the detected surface condition.

System and method for controlling a vehicle steering system

Techniques and examples pertaining to vehicle odometry using one or more radars disposed on a vehicle are described. A method for radar odometry may involve receiving, by a processor from the radars, measurement data of stationary objects and moving objects that are located in an environment a vehicle is traversing. The method may also involve performing, by the processor, a random sample consensus (RANSAC) calculation to select the measurement data of the stationary objects and disregard the measurement data of the moving objects. The method may also involve calculating, by the processor, one or more dynamic variables of the vehicle based on the measurement data of the stationary objects. The proposed method processes the measurement data of the stationary objects with one single RANSAC calculation and one least squares problem solving, thereby greatly reducing computation cost and time as well as latency in operation for providing the vehicle odometry.

Radar odometry for vehicle

A hitch assist system and method are provided herein. An imager captures one or more images of a trailer having a hitch coupler. A user-input device is provided for specifying a location of the hitch coupler in the one or more captured images. A controller generates an image patch around the user-specified location, segments the image patch to identify the hitch coupler, and estimates a trailer height based on a height and projective geometry of the imager.

Growing community of inventors

Mountain View, CA, United States of America

Mohsen Lakehal-ayat