Method and system for controlling safety of ego and social objects

Patent metadata

Field	Value
Publication number	US-11364936-B2
Application number	US-202016803386-A
Country	US
Kind code	B2
Filing date	Feb 27, 2020
Priority date	Feb 28, 2019
Publication date	Jun 21, 2022
Grant date	Jun 21, 2022

Abstract

A method or system for controlling safety of both an ego vehicle and social objects in an environment of the ego vehicle, comprising: receiving data representative of at least one social object and determining a current state of the ego vehicle based on sensor data; predicting an ego safety value corresponding to the ego vehicle, for each possible behavior action in a set of possible behavior actions, based on the current state; predicting a social safety value corresponding to the at least one social object in the environment of the ego vehicle, based on the current state, for each possible behavior action; and selecting a next behavior action for the ego vehicle, based on the ego safety values, the social safety values, and one or more target objectives for the ego vehicle.

First claim

The invention claimed is:

1. A method for controlling an ego vehicle in an environment that the ego vehicle and a social object are is-operating in, the method comprising:
determining a current state of the ego vehicle based on sensor data obtained from a plurality of sensors of the ego vehicle, the current state including data about (i) operating state of the ego vehicle and (ii) obstacles in the environment including the social object;
predicting, based on the current state, for each possible behavior action of the ego vehicle in a set of possible behavior actions of the ego vehicle, an ego safety value indicating a probability that an ego safety zone of the ego vehicle will be free of obstacles if the possible behavior action is performed by the ego vehicle, wherein the ego safety zone comprises a physical space that includes and extends beyond the ego vehicle in a direction of travel of the ego vehicle;
predicting, based on the current state, for each possible behavior action of the ego vehicle in the set of possible behavior actions, a social safety value corresponding to the social object, the social safety value for each possible behavior action indicating a probability that the ego vehicle will not be located in a social safety zone of the social object if the possible behavior action is performed by the ego vehicle, wherein the social safety zone comprises a physical space that includes and extends beyond the social object in a direction of travel of the social object;
selecting, based on the ego safety values, the social safety values, and one or more target objectives for the ego vehicle, a next behavior action for the ego vehicle to keep the obstacles out of the ego safety zone and to keep the ego vehicle out of the social safety zone of the social object;
determining a planned trajectory based on the selected next behavior action for the ego vehicle; and
controlling a drive control system of the ego vehicle to implement the planned trajectory.

2. The method of claim 1 wherein the data about obstacles in the environment comprises data representative of a plurality of further social objects in addition to the soda object, the method including, for each possible behavior action, predicting a respective social safety value for each of the further social objects, each respective social safety value for each further social object indicating a probability that the ego vehicle will not he located in a respective soda safety zone of the respective further social object if the possible behavior action is performed by the ego vehicle.

3. The method of claim 2 wherein determining the current state comprises determining a velocity and direction of the ego vehicle, and a velocity, direction and position of the social object and each of the plurality of further social objects.

4. The method of claim 2 wherein predicting the ego safety value for each possible behavior action is performed by a general value function (GVF) implemented by a first trained neural network; and predicting the social safety value for each possible behavior action for the social object and each of the plurality of further social objects is performed by a further GVF implemented by a second trained neural network, wherein the first and second trained neural network have each been trained to incorporate a target policy of the ego vehicle.

5. The method of claim 1 wherein a size of one or both of the ego safety zone and the social safety zone is based on the current state.

6. The method of claim 1 wherein, for each possible behavior action, the social safety value corresponds to a plurality of social objects in the environment of the ego vehicle, the social safety value indicating a probability that the ego vehicle will not be located in a respective social safety zone of any of the social objects if the possible behavior action is performed by the ego vehicle.

7. The method of claim 1 wherein selecting the next behavior action for the ego vehicle comprises:
performing fuzzification of the ego safety value and the social safety value predicted for each of the possible behavior actions by mapping each of the ego safety values and the social safety values to a respective truth value;
applying fuzzy inference on the truth values to generate a goal fuzzy set; and
defuzzifying the goal fuzzy set to select the next behavior action for the ego vehicle.

8. The method of claim 1 further comprising, for each possible behavior action, predicting, based on the current state, an ego comfort value corresponding to an acceleration of the ego vehicle, and wherein selecting the next behavior action for the ego vehicle is also based on the ego comfort values predicted for each possible behavior action.

9. A system for controlling an ego vehicle in an environment that the ego vehicle and a social object are operating in, the system comprising:
one or more processor systems;
a memory storing machine-executable instructions which, when executed by the one or more processor systems, cause the system to:
determine a current state based on sensor data obtained from a plurality of sensors of the ego vehicle, the current state including data about (i) operating state of the ego vehicle and (ii) obstacles in an environment of the ego vehicle, including a social object;
predict, based on the current state, for each possible behavior action of the ego vehicle in a set of possible behavior actions, an ego safety value, indicating a probability that an ego safety zone of the ego vehicle will be free of obstacles if the possible behavior action is performed by the ego vehicle, wherein the ego safety zone comprises a physical space that includes and extends beyond the eco vehicle in a direction of travel of the ego vehicle;
predict, based on the current state, for each of the possible behavior actions of the ego vehicle in the set of possible behavior actions, a social safety value corresponding to at the social object, the social safety value for each possible behavior action indicating a probability that the ego vehicle will not be located in a social safety zone of the social object if the possible behavior action is performed by the ego vehicle, wherein the social safety zone comprises a physical space that includes and extends beyond the social object in a direction of travel of the social object;
select, based on the ego safety values, the social safety values predicted for each of the possible behavior actions, and one or more target objectives, a next behavior action for the ego vehicle to keep the obstacles out of the ego-vehicle's safety zone and keep the ego-vehicle out of the safety zone of the social object;
determine a planned trajectory based on the selected next behavior action for the ego vehicle; and
send the planned trajectory to a drive control system of the ego vehicle for implementation of the planned trajectory.

10. The system of claim 9 wherein the data about obstacles includes data representative of a plurality of further social obstacles in addition to the social obstacle, and the machine-executable instructions, when executed by the one or more processor systems, cause the system to predict, for each possible behavior action, a respective social safety value for each of the plurality of further social objects in the environment of the ego vehicle, each respective social safety value for each further social object indicating a probability that the ego vehicle will not be located in a respective social safety zone of the respective further social object if the possible behavior action is performed by the ego vehicle.

11. The system of claim 10 wherein the current state comprises velocity and direction of the ego vehicle, an

Assignees

• Graves Daniel Mark
• Huawei Tech Co Ltd

Inventors

• Graves Daniel Mark

Classifications

• G06N3/045

  Combinations of networks · CPC title

• G06N3/092

  Reinforcement learning · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• B60W60/0017

  of other traffic participants · CPC title

• B60W60/0011

  involving control alternatives for a single driving scenario, e.g. planning several paths to avoid obstacles · CPC title