Abstract: A computer, including a processor and a memory, the memory including instructions to be executed by the processor to input a fisheye image to a vector quantized variational autoencoder. The vector quantized variational autoencoder can encode the fisheye image to first latent variables based on an encoder. The vector quantized variational autoencoder can quantize the first latent variables to generate second latent variables based on a dictionary of embeddings. The vector quantized variational autoencoder can decode the second latent variables to a rectified rectilinear image using a decoder and output the rectified rectilinear image.

Abstract: A computer, including a processor and a memory, the memory including instructions to be executed by the processor to input a fisheye image to a vector quantized variational autoencoder. The vector quantized variational autoencoder can encode the fisheye image to first latent variables based on an encoder. The vector quantized variational autoencoder can quantize the first latent variables to generate second latent variables based on a dictionary of embeddings. The vector quantized variational autoencoder can decode the second latent variables to a rectified rectilinear image using a decoder and output the rectified rectilinear image.

Abstract: Based on an image from a stationary sensor, a marker displayed on one or more digital displays on a vehicle is detected. A first location and a first orientation of the marker in a coordinate system is determined by analyzing on pixels in the image. Based on a stationary sensor location and orientation from a map, a second location and a second orientation of the vehicle in the coordinate system is determined.

Abstract: Based on an image from a stationary sensor, a marker displayed on one or more digital displays on a vehicle is detected. A first location and a first orientation of the marker in a coordinate system is determined by analyzing on pixels in the image. Based on a stationary sensor location and orientation from a map, a second location and a second orientation of the vehicle in the coordinate system is determined.

Abstract: A vehicle is disclosed that uses data from different types of on-board sensors to determine whether meteorological precipitation is failing near the vehicle. The vehicle may include an on-board camera capturing image data and an on-board accelerometer capturing accelerometer data. The image data may characterize an area in front of or behind the vehicle. The accelerometer data may characterize vibrations of a windshield of the vehicle. An artificial neural network may run on computer hardware carried on-board the vehicle. The artificial neural network may be trained to classify meteorological precipitation in an environment of the vehicle using the image data and the accelerometer data as inputs. The classifications of the artificial neural network may be used to control one or more functions of the vehicle such as windshield-wiper speed, traction-control settings, or the like.

Abstract: Techniques and implementations pertaining to detection of moving objects, such as pedestrians, when a vehicle moves in a rearward direction are described. A method may involve identifying a region of interest when a vehicle moves in a rearward direction. The method may involve detecting a moving object in the region of interest. The method may also involve determining whether a collision with the moving object by the vehicle moving in the rearward direction is likely. The method may further involve providing a human-perceivable signal responsive to a determination that the collision is likely.

Abstract: The present invention extends to methods, systems, and computer program products for detecting and responding to emergency vehicles in a roadway. Aspects of the invention can be used to detect emergency vehicles and properly yield to emergency vehicles depending on roadway configuration. A vehicle includes a plurality of sensors. The vehicle also includes vehicle to vehicle (V2V) communication capabilities and has access to map data. Sensor data from the plurality of sensors along with map data is provided as input to a neural network (either in the vehicle or in the cloud). Based on sensor data, the neural network detects when one or more emergency vehicles are approaching the vehicle. From a roadway configuration, a vehicle can use the plurality of sensors to automatically (and safely) yield to detected emergency vehicle(s). Automatically yielding can include one or more of: slowing down, changing lanes, stopping, etc.

Abstract: A vehicle is disclosed that uses data from different types of on-board sensors to determine whether meteorological precipitation is failing near the vehicle. The vehicle may include an on-board camera capturing image data and an on-board accelerometer capturing accelerometer data. The image data may characterize an area in front of or behind the vehicle. The accelerometer data may characterize vibrations of a windshield of the vehicle. An artificial neural network may run on computer hardware carried on-board the vehicle. The artificial neural network may be trained to classify meteorological precipitation in an environment of the vehicle using the image data and the accelerometer data as inputs. The classifications of the artificial neural network may be used to control one or more functions of the vehicle such as windshield-wiper speed, traction-control settings, or the like.

Abstract: Techniques and implementations pertaining to detection of moving objects, such as pedestrians, when a vehicle moves in a rearward direction are described. A method may involve identifying a region of interest when a vehicle moves in a rearward direction. The method may involve detecting a moving object in the region of interest. The method may also involve determining whether a collision with the moving object by the vehicle moving in the rearward direction is likely. The method may further involve providing a human-perceivable signal responsive to a determination that the collision is likely.