18481164. LEARNING LATENT-SPACE BARRIER FUNCTIONS FROM SAFE DEMONSTRATIONS simplified abstract (TOYOTA RESEARCH INSTITUTE, INC.)
Contents
LEARNING LATENT-SPACE BARRIER FUNCTIONS FROM SAFE DEMONSTRATIONS
Organization Name
TOYOTA RESEARCH INSTITUTE, INC.
Inventor(s)
Fernando Castaneda Garcia-rozas of Berkeley CA (US)
Haruki Nishimura of Sunnydale CA (US)
Rowan Mcallister of San Jose CA (US)
Adrien David Gaidon of Mountain View CA (US)
LEARNING LATENT-SPACE BARRIER FUNCTIONS FROM SAFE DEMONSTRATIONS - A simplified explanation of the abstract
This abstract first appeared for US patent application 18481164 titled 'LEARNING LATENT-SPACE BARRIER FUNCTIONS FROM SAFE DEMONSTRATIONS
The abstract describes a method for a task-agnostic policy filter control system, involving encoding current observations, previous latent states, and previous actions to output a new latent state. The method utilizes a neural ordinary differential equations (ODE) module to compute learned latent state-space dynamic models, an in-distribution barrier function (iDBF) model to infer iDBF values, and computes a reference control input and current action based on the learned models to keep the system in-distribution with respect to safe demonstrations.
- Encoding current observations, previous latent states, and previous actions to output a new latent state
- Using a neural ODE module to compute learned latent state-space dynamic models
- Inferring iDBF values using an in-distribution barrier function model
- Computing reference control inputs and current actions based on learned models
- Keeping the system in-distribution with respect to safe demonstrations
Potential Applications: - Autonomous vehicles - Robotics - Industrial automation - Healthcare monitoring systems
Problems Solved: - Ensuring safe and reliable control systems - Adapting to changing environments - Maintaining system stability
Benefits: - Improved safety - Enhanced adaptability - Increased efficiency
Commercial Applications: Title: "Enhancing Control Systems for Autonomous Vehicles and Robotics" This technology can be applied in autonomous vehicles, robotics, industrial automation, and healthcare monitoring systems, improving safety, adaptability, and efficiency in various industries.
Questions about the technology: 1. How does the method ensure the system remains in-distribution with respect to safe demonstrations? - The method computes reference control inputs and current actions based on learned models to maintain system safety. 2. What are the key components of the neural ODE module in this method? - The neural ODE module is used to compute learned latent state-space dynamic models for the new latent state.
Original Abstract Submitted
A method for a task-agnostic policy filter control system is described. The method includes encoding a current observation, a previous latent state, and a previous action to output a new latent state. The method also includes computing, by a neural ordinary differential equations (ODE) module, learned latent state-space dynamic models for the new latent state. The method further includes inferring, by an in-distribution barrier function (iDBF) model, an iDBF value in response to the new latent state. The method also includes computing, based on the learned latent state-space dynamic models, the iDBF value and a reference control input for a current timestep, and a current action. The current action keeps the task-agnostic policy filter control system in-distribution with respect to an offline-collected dataset of safe demonstrations.
