Determining drivable free-space for autonomous vehicles

Abstract

In various examples, sensor data may be received that represents a field of view of a sensor of a vehicle located in a physical environment. The sensor data may be applied to a machine learning modelthat computes both a set of boundary points that correspond to a boundary dividing drivable free-space from non-drivable space in the physical environment and class labels for boundary points of the set of boundary points that correspond to the boundary. Locations within the physical environment may be determined from the set of boundary points represented by the sensor data, and the vehicle may be controlled through the physical environment within the drivable free-space using the locations and the class labels.

Description

Background technique [0001] An autonomous driving system can control an autonomous vehicle without human supervision while achieving an acceptable level of safety. This may require autonomous driving systems capable of at least the functional performance of an attentive human driver utilizing perception and action systems with an incredible ability to recognize and react to moving and static obstacles in complex environments . To achieve this, areas of the accessible environment (e.g., drivable free space) can be determined, as this information may be useful for autonomous driving systems and / or advanced driver assistance systems (ADAS) when planning maneuvers and / or navigation decisions it works. [0002] Some traditional methods of determining drivable free space use vision-based techniques using deep artificial neural networks (DNNs). For example, these traditional approaches use DNNs such as convolutional neural networks (CNNs) to perform semantic segmentation (e.g., pi...

AI Technical Summary

Problems solved by technology

For example, without the context of a class of boundaries provided by CNNs—e.g., dynamic boundaries such as humans, as opposed to static boundaries such as curbs—autonomous driving systems may not be able to accurately predict the drivable free space in a way that the autonomous vehicle can be safely controlled

Images