20240010196. LEARNING AUTONOMOUS VEHICLE SAFETY CONCEPTS FROM DEMONSTRATIONS simplified abstract (NVIDIA Corporation)

From WikiPatents
Jump to navigation Jump to search

LEARNING AUTONOMOUS VEHICLE SAFETY CONCEPTS FROM DEMONSTRATIONS

Organization Name

NVIDIA Corporation

Inventor(s)

Karen Yan Ming Leung of Los Altos CA (US)

Sushant Veer of Sunnyvale CA (US)

Edward Fu Schmerling of Los Altos CA (US)

Marco Pavone of Stanford CA (US)

LEARNING AUTONOMOUS VEHICLE SAFETY CONCEPTS FROM DEMONSTRATIONS - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240010196 titled 'LEARNING AUTONOMOUS VEHICLE SAFETY CONCEPTS FROM DEMONSTRATIONS

Simplified Explanation

The abstract of this patent application describes a method for learning control policies for agents navigating through an environment. The control policies are learned from demonstrations that capture the joint states of the entities involved. The learned control policy maps joint states to control actions, where the joint states are between agents and the control actions are performed by at least one of the agents. The control policy defines mappings as control invariant sets of joint states and control actions. The control policy can also determine functions that compute the likelihood of collision between entities based on their joint state. This information can be used to evaluate the current and potential states of the environment and determine control operations for a machine, such as a vehicle.

  • Control policies for controlling agents can be learned from demonstrations capturing joint states of entities navigating through the environment.
  • The control policy maps joint states to control actions, where the joint states are between agents and the control actions are performed by at least one of the agents.
  • The control policy defines mappings as control invariant sets of joint states and control actions.
  • The control policy can determine functions that compute the likelihood of collision between entities based on their joint state.
  • The output of the functions can be used to evaluate the current and potential states of the environment.
  • Control operations for a machine, such as a vehicle, can be determined based on the evaluation of the environment states.

Potential applications of this technology:

  • Autonomous vehicles: The learned control policies can be applied to autonomous vehicles to navigate through complex environments and avoid collisions.
  • Robotics: The control policies can be used in robotic systems to control the movements of multiple robots operating in the same environment.
  • Traffic management: The control policies can be utilized in traffic management systems to optimize the flow of vehicles and prevent accidents.

Problems solved by this technology:

  • Collision avoidance: The control policies and functions for computing collision likelihood help in preventing collisions between entities operating in the environment.
  • Efficient navigation: The learned control policies enable agents to navigate through the environment in an efficient and safe manner.
  • Complex environment handling: The technology allows for the control of agents in complex environments with multiple entities, where traditional control methods may be insufficient.

Benefits of this technology:

  • Improved safety: By learning control policies and evaluating collision likelihood, the technology enhances safety by preventing collisions between entities.
  • Increased efficiency: The control policies enable agents to navigate through the environment efficiently, optimizing their movements and reducing unnecessary actions.
  • Adaptability: The technology can adapt to different environments and entities, allowing for flexible control operations based on the specific context.


Original Abstract Submitted

in various examples, control policies for controlling agents may be learned from demonstrations capturing joint states of entities navigating through the environment. a control policy may be learned mapping joint states to control actions, where the joint states are between agents, and the control actions are of at least one of the agents. the control policy may be learned to define the mappings as control invariant sets of the joint sates and the control actions. the control policy may be used to determine one or more functions that compute, based at least on a joint state between entities, output indicating a likelihood of collision between the entities operating in accordance with the control policy. using the output, current and/or potential states of the environment may be evaluated to determine control operations for a machine, such as a vehicle.

Retrieved from "http://wikipatents.org/index.php?title=20240010196._LEARNING_AUTONOMOUS_VEHICLE_SAFETY_CONCEPTS_FROM_DEMONSTRATIONS_simplified_abstract_(NVIDIA_Corporation)&oldid=27470"