Methods for risk management for autonomous devices and related node

1 . A method performed by a risk management node, the method comprising:

generating a scene graph structure of an environment based on sensor data, wherein the scene graph structure includes at least one object, a relationship of the least one object with an autonomous device and the environment, and one or more environment parameters;

determining state parameters from the scene graph structure of the environment that includes the at least one object, the autonomous device, and a set of safety zones for the autonomous device relative to the at least one object, wherein determining the state parameters comprises converting each of the one or more environment parameters to a discrete state parameter;

determining a reward value for the autonomous device based on evaluating a risk of a hazard with the least one object based on the determined state parameters and current location and current speed of the autonomous device relative to a safety zone from the set of safety zones;

determining a control parameter for controlling action of the autonomous device based on the determined reward value; and

initiating sending the control parameter to the autonomous device to control action of the autonomous device,

wherein the control parameter is dynamically adapted to reduce the risk of hazard with the at least one object based on reinforcement learning feedback from the reward value,

wherein the discrete state parameter comprises:

a current direction of the autonomous device;

a current speed of the autonomous device;

a current location of the autonomous device;

a distance of the at least one obstacle from a safety zone in the set of safety zones for the autonomous device;

a direction of the at least one object relative to a surface of the autonomous device; and

a risk value for the at least one object based on a classification of the at least one object.

2 . The method of claim 1 , wherein the risk management node is onboard the autonomous device.

3 . The method of claim 1 , wherein the state parameters are determined from inputting to the risk management node the one or more environment parameters from the scene graph structure.

4 . A computer program product comprising:

non-transitory computer readable medium storing instructions, when executed on at least one processor causes the at least one processor to carry out a method according to claim 1 .

5 . The method of claim 1 , wherein the scene graph structure includes a first object, a second object, and a relationship of the first object with the second object, the autonomous device, and the environment.

6 . The method of claim 1 , wherein generating the scene graph structure comprises extracting the at least one object from the sensor data using a computer vision system.

7 . A risk management node, the risk management node comprising:

at least one processor; and

at least one memory connected to the at least one processor and storing program code that is executed by the at least one processor to perform operations comprising:

generating a scene graph structure of an environment based on sensor data, wherein the scene graph structure includes at least one object, a relationship of the least one object with an autonomous device and the environment, and one or more environment parameters;

determining state parameters from the scene graph structure of the environment that includes the at least one object, the autonomous device, and a set of safety zones for the autonomous device relative to the at least one object, wherein determining the state parameters comprises converting each of the one or more environment parameters to a discrete state parameter;

determining a reward value for the autonomous device based on evaluating a risk of a hazard with the least one object based on the determined state parameters and current location and current speed of the autonomous device relative to a safety zone from the set of safety zones;

determining a control parameter for controlling action of the autonomous device based on the determined reward value; and

initiating sending the control parameter to the autonomous device to control action of the autonomous device,

wherein the control parameter is dynamically adapted to reduce the risk of hazard with the at least one object based on reinforcement learning feedback from the reward value,

wherein the discrete state parameter comprises:

a current direction of the autonomous device;

a current speed of the autonomous device;

a current location of the autonomous device;

a distance of the at least one obstacle from a safety zone in the set of safety zones for the autonomous device;

a direction of the at least one object relative to a surface of the autonomous device; and

a risk value for the at least one object based on a classification of the at least one object.

8 . The risk management node of claim 7 , wherein the at least one object comprises a static item, a dynamic item, or a human.

9 . The risk management node of claim 7 , further comprising:

repeating the determining the state parameters, the determining the reward value, the determining the control parameter, and the initiating sending the control parameter to control action of the autonomous device.

10 . The risk management node of claim 7 , wherein the classification of the object comprises an attribute parameter identifying the at least one object as comprising a human, an infrastructure, another autonomous device, or a vehicle.

11 . The risk management node of claim 7 , wherein the set of safety zones comprise a range of safety zones, wherein each safety zone in the range has a different distance from the autonomous device and the autonomous device has a different speed within each safety zone within the range of safety zones.

12 . The risk management node of claim 7 , wherein the reward value comprises a defined numerical value based on the evaluated risk of hazard with the at least one object.

13 . The risk management node of claim 7 , wherein the control parameter is a speed of at least one actuator of the autonomous device.

14 . The risk management node of claim 7 , wherein the control parameter is an angle of at least one actuator of the autonomous device.

15 . The risk management node of claim 7 , wherein the one or more environment parameters comprise at least one of:

a distance of the at least one object from a surface of the autonomous device;

an orientation of a surface of the at least one object from the autonomous device;

a direction of the at least one object from a surface of the autonomous device;

a velocity of the at least one object;

a width dimension of the at least one object;

a length dimension of the at least one object; and

a height dimension of the least one object.

16 . The risk management node of claim 7 , wherein the risk management node is onboard the autonomous device.

17 . The risk management node of claim 7 , wherein the state parameters are determined from inputting to the risk management node each of the one or more environment parameters from the scene graph structure.

18 . The risk management node of claim 7 , further comprising:

sending the control parameter to a controller for the autonomous device for application to a trajectory for the autonomous device.

19 . The risk management node of claim 7 , wherein the risk value for the at least one object based on the classification of the at least one object is input to the risk management node from a risk analysis module that assigns the risk value.