Online occlusion prediction for autonomous vehicles

A vehicle, the vehicle comprising:

• one or more sensors sensing an area;
• and
• a compute system to:
• receive sensor data from the one or more sensors of the vehicle sensing the area;
• detect an object based on the sensor data;
• determine object information of the detected object based on the sensor data, the object information including location coordinates;
• translate the location coordinates of the detected object from a local reference frame of the vehicle to a common reference frame shared by the vehicle and a further vehicle;
• compress, using an encoder implemented on the compute system, sensor data corresponding to the detected object into an array, wherein the array comprises feature embeddings capturing features of the detected object;
• transmit the object information, the array, and a timestamp associated with the object information and the array to a system external to the vehicle, wherein the object information comprises the translated location coordinates;
• receive a second array which comprises feature embeddings capturing features of a second object, wherein the second object is in the area and is outside a field of perceivable area of the one or more sensors of the vehicle;
• incorporate the object information, the array, the timestamp, and the second array to a temporal map;
• and
• control the vehicle based on the temporal map.

One or more non-transitory computer-readable media storing instructions that, when executed by one or more processors, cause the one or more processors to:

• receive sensor data from one or more sensors of a vehicle sensing an area;
• detect an object based on the sensor data;
• determine object information of the detected object based on the sensor data, the object information including location coordinates;
• translate the location coordinates of the detected object from a local reference frame of the vehicle to a common reference frame shared by the vehicle and a further vehicle;
• compress sensor data corresponding to the detected object into an array, wherein the array comprises feature embeddings capturing features of the detected object;
• transmit the object information, the array, and a timestamp associated with the object information and the array to an external system, wherein the object information comprises the translated location coordinates;
• receive a second array which comprises feature embeddings capturing features of a second object, wherein the second object is in the area and is outside a field of perceivable area of the one or more sensors of the vehicle;
• incorporate the object information, the array, the timestamp, and the second array to a temporal map;
• and
• control the vehicle based on the temporal map.