Controlling autonomous vehicles using safe arrival times

Abstract

In various examples, sensor data representative of a field of view of at least one sensor of a vehicle in an environment is received from the at least one sensor. Based at least in part on the sensordata, parameters of an object located in the environment are determined. Trajectories of the object are modeled toward target positions based at least in part on the parameters of the object. From thetrajectories, safe time intervals (and / or safe arrival times) over which the vehicle occupying the plurality of target positions would not result in a collision with the object are computed. Based atleast in part on the safe time intervals (and / or safe arrival times) and a position of the vehicle in the environment a trajectory for the vehicle may be generated and / or analyzed.

Description

Background technique [0001] To operate safely, an autonomous vehicle should be able to traverse a path through the environment without colliding with other objects (e.g., vehicles, people, trees, animals, buildings, etc.). To achieve this, an autonomous vehicle should consider other objects when planning a path through the environment. Traditional approaches for path planning for autonomous vehicles include generating a heat map of the environment around the autonomous vehicle at a given time, and using the heat map to plan a path for the vehicle. To generate the heat map, locations of other objects may be determined, and a probability of the vehicle colliding with the other objects at a particular location may be determined based at least in part on the object locations. Probabilities can be represented by heatmaps and other information such as uncertainty in object locations can be taken into account. [0002] However, the conventional methods described above do not take i...

AI Technical Summary

Problems solved by technology

[0002] However, the traditional methods described above do not take into account potential changes in the motion or state (e.g., velocity, velocity, acceleration, etc.) of objects in the environment at one or more points along the path.

As a result, paths generated for autonomous vehicles using these traditional methods may not fully account for possible collisions between vehicles and objects

This may present some challenges that may make systems that rely on these methods unsafe

For example, using these traditional methods, paths may quickly become less and less safe as they are followed for longer periods of time, which may require the system to use short paths, and frequently update or generate short paths

As a result, systems relying on these traditional approaches may not exhibit sufficient foresight to allow the system to be informed as fully autonomous driving would expect

Additionally, because paths may need to be updated and / or generated so frequently, significant computational resources may be required

Images